Copper is widely used in drinking water distribution systems due to its relatively low cost and favorable mechanical properties. However, copper corrosion may generate copper concentrations exceeding the thresholds prescribed by international drinking water standards. In-situ measurements performed in an actual system found that the copper mass released under flowing water conditions (pipe flushing) was greater than the copper mass release estimated considering only the mass of copper in the pipe's bulk water before the tap is opened. This work presents in-situ and laboratory results of copper release into the water and its dependence on biofilm structure, solid-liquid interface properties, and the pipe flow regime (laminar vs. transition flow). The results of this work highlight the necessity to incorporate the hydrodynamic effects in the analysis of corrosion and corrosion by-products release into drinking water piping systems. Initial modeling efforts are also presented.

INTRODUCTION

For decades, copper has been used for household drinking water systems around the world. This is probably due to its corrosion resistance and scale-forming properties. However, under specific physical–chemical–biological water quality and operational conditions, it is affected by corrosion (Callot et al. 1987). This leads to the release of copper to the water, making it unsafe for human consumption. Microbially influenced corrosion (MIC) has been identified as a severe type of corrosion that has caused costly failures around the world (Critchley et al. 2004). Traditional models of MIC in copper pipe systems do not completely describe the interaction between the biofilm and corrosion by-products under flow-stagnation cycles.

It is not clear how fluid flow affects the concentration of copper released. Werner (1995) observed that, for constant flow conditions (24 hours), the copper released once the flow is stopped reaches a maximum around 3–5 hours after the flow was stopped. For conditions where the pipe was subject to 1 hour of flow, the copper concentration during stagnation increases and reaches a maximum. These experiments suggest that copper dissolves under flow and stagnation conditions, but only during stagnation are copper ions accumulated until reaching equilibrium with the solid by-products formed on the pipe walls.

In the case of MIC (Lehtola et al. 2006), there is evidence that increasing the flow might induce the growth of biofilm, also there is an increase of bacteria and copper release into the water, probably related to the detachment of corrosion by-products and bacteria from the pipe's wall. Thus, the objective of this article is to present experimental evidence and modeling experiments for highlighting the necessity to incorporate the hydrodynamic effects in the analysis of copper contamination due to corrosion of drinking water piping systems, especially in cases of MIC.

MATERIALS AND METHODS

Our study was developed in both a household drinking water system affected by MIC, and laboratory experiments consisting of copper pipes supplied by synthetic water. Tested pipes were flushed at a constant laminar flow rate according to the methodology presented in Calle et al. (2007) and Vargas et al. (2010).

The household water system consisted of a well connected to a PVC pipe followed by a 1 m long copper pipe with the same characteristics as that used in the laboratory experiments. This system was in operation for 2 years before the analysis. The UV disinfection system stopped working a year before the experiments were performed. During 10 successive days, 15 samples were taken for each flow experiment after stagnation of 10 hours. Flow was kept at a constant rate of 0.48 L/min, which resulted in laminar conditions within the pipe.

The laboratory tests were conducted on a 1 m long copper pipe with an internal diameter of 1.95 cm and 0.3 L of volume. Pipes were previously aged by water changes three times per week (Monday, Wednesday and Friday) (Boulay & Edwards 2001) under biotic conditions with water extracted from a real system affected by MIC (Calle et al. 2007). After 4 weeks of ageing and 8 hours of stagnation, pipes were flushed at a flow rate of 0.24 L/min at 25 °C which is representative of domestic pipe systems under laminar conditions (Re = 280.44). During flushing experiments, sequential water samples of 15 mL for the first 0.6 L and 100 mL for the rest, were taken from the tap to determine dissolved and total copper concentrations until approximately 6 L of water had been flushed through. An ICP spectrometer (Thermo Electron iCAP 6300) was used to measure copper released into the water. The mass of copper released was estimated by integrating the copper concentration evolution versus the volume of water extracted from the pipe.

Three principal tasks were performed for field and laboratory experiments:

  • (a) Sampling and in-situ wet chemistry analyses (e.g. pH, conductivity, alkalinity, hardness, and dissolved organic carbon).

  • (b) Characterization of surface corrosion by-products. Scanning electron microscopy (SEM) and energy dispersive spectroscopy (EDS) were used to study the morphology of the biofilm matrix formed on the inner surface of copper pipes affected by MIC. Several 1 × 1 cm coupons were cut from laboratory and field pipes. The coupons were treated with critical point drying (Carr et al. 1996; Schadler et al. 2008) and coated with a thin gold film. A LEO 1420VP scanning electron microscope coupled to an Oxford 7424 solid-state detector was used to obtain the micrographs and an estimation of the elemental composition.

  • (c) Mathematical modelling of hydrodynamics and transport. A hydrodynamic mathematical model for flushing in laminar flow was developed using the Navier-Stokes and advection-diffusion equations in cylindrical coordinates (Pizarro et al. 2014). This model considers: (i) an initial concentration profile of total copper in any section of the pipe at the end of the stagnation period; and (ii) a model of copper transport by advection during flushing.

RESULTS AND DISCUSSION

We characterized stagnation-flushing cycles in the pipe subject to corrosion in the presence of a microbial biofilm. Figure 1 (left and right panels) presents the results for the flushing experiments performed in the field and laboratory. The graph presents the concentration of Cu vs. the volume of water extracted from the pipe. Field and laboratory results show similar patterns, where the concentration of Cu is higher at the beginning of the flushing. The initial concentration corresponds to the concentration within the pipe after stagnation. After the initial volume is extracted, the Cu concentration decreases to a stable value as the flushing proceeds. The differences observed are principally due to the ageing time that these pipes had (2 years and 4 weeks). In a previous study aimed to analyze and compare the composition of bacterial biofilms from both field (same system) and laboratory (tests using the same pipe material, but no flow) conditions, we observed not only differences in biofilm structure, but also dissimilar bacterial composition, suggesting a microbial succession over time. The community developed in the mature biofilm present in the field seems to be more adapted to high copper concentrations (Pavissich et al. 2010).
Figure 1

Copper release curves for field system (left panel) and laboratory system (right panel).

Figure 1

Copper release curves for field system (left panel) and laboratory system (right panel).

Copper corrosion studies are usually centered on understanding the processes during stagnation periods and the copper release determined by these processes. Considering that the initial concentration measured with the flushing experiment is the concentration of copper within the pipe, we can compute the mass of copper in the bulk liquid as this concentration multiplied by the volume of water within the pipe. Also, by integrating the curves of copper release vs. water extracted from the pipe, we can compute the mass of copper released for the first 1, 3, and 6 liters extracted from the pipes.

The results of the microscopic characterization of the inner surface of the pipe showed it covered by a biological matrix formed by a patchwork of bacteria, EPS (extracellular polymeric substances) and inorganic corrosion by-products (Figure 2(a) and 2(b)). Although EDS analysis revealed similar elemental composition of the corrosion by-products formed on the metallic surface of both pipes (C = 12 wt%; O = 7.7 wt%; Cu = 71.5 wt%; Si = 7.9 wt% and C = 6.6 wt%; O = 2.3 wt%; Cu = 91.1 wt%, respectively), probably indicating the presence of copper carbonates as malachite; differences in this patchwork structure were observed. The field pipe shows a mature biofilm, increasing heterogeneity and changing the surface morphology. These differences observed on the surface were also measured in the water. The mass of copper released during flushing experiments was about three (laboratory) to nine (field) times the mass of copper in the bulk water before the tap was opened (plug flow assumption, Figure 2(c)). This difference is probably due to ageing (i.e. biofilm development) (Pavissich et al. 2010), which plays an important role in feature-rich surface development.
Figure 2

(a) SEM image of the inner surface of the copper pipe tested in the laboratory (4 weeks) shows the close interactions between bacteria, EPS and nanosized mineralizations. (b) SEM image of the mature biofilm formed in the pipe extracted from the field after 2 years of operation. (c) Comparison of mass of copper released for different pipe ages and volume of water extracted.

Figure 2

(a) SEM image of the inner surface of the copper pipe tested in the laboratory (4 weeks) shows the close interactions between bacteria, EPS and nanosized mineralizations. (b) SEM image of the mature biofilm formed in the pipe extracted from the field after 2 years of operation. (c) Comparison of mass of copper released for different pipe ages and volume of water extracted.

The hypothesis is that this complex surface (biofilm, EPS, and corrosion by-products) plays an active role liberating dissolved and nano-particulated copper into the bulk liquid. Thus, possible mechanisms that would explain the higher copper release in pipes with flow might be the release of nanoparticles from the pipe's wall (Pavissich et al. 2010; Olivares et al. 2014), and desorption of labile copper complexed with the biofilm (Pizarro et al. 2014) (Figure 3).
Figure 3

Schematic representation of the possible phenomena occurring during flow conditions.

Figure 3

Schematic representation of the possible phenomena occurring during flow conditions.

To explain this behaviour a mathematical model was developed where, in order to reproduce the observed copper release behaviour, the concentration of copper close to the surface needs to be increased (Calle et al. 2007; Pizarro et al. 2014). This hydrodynamic effect on copper release was observed for both laboratory and field systems.

The mathematical model developed considered the following processes:

  • (a) Growth and decay of the biofilm on the inner surface of the pipe.

  • (b) Transport of chemical species, by diffusion, during stagnation time.

  • (c) Transport of chemical species, by convection, during flow conditions.

  • (d) Speciation of copper.

  • (e) Release of copper ions into the liquid.

The parameters used are described elsewhere (Pizarro et al. 2014). The biofilm development was modelled for a long period of time (350 days), using a discrete cellular automaton model (Pizarro et al. 2001, 2004, 2005). A representative structure is depicted in Figure 4, upper panel, for day 238 of the simulation using a mesh size of 10 μm. The lower panel of the figure presents the concentration of complexed copper within the biofilm after a stagnation period of 10 h, given a rate of copper release due to corrosion and the complexation capacity estimated experimentally.
Figure 4

Upper panel: Modelled biofilm structure for stagnation-flow simulation. Bottom panel: Spatial distribution of dissolved and complexed copper after 10 h stagnation.

Figure 4

Upper panel: Modelled biofilm structure for stagnation-flow simulation. Bottom panel: Spatial distribution of dissolved and complexed copper after 10 h stagnation.

Using the hydrodynamic model for a laminar flow and the initial concentration of copper associated with the presence of biofilm on the surface of the pipe, we were able to model the release curve of the concentration of copper as the flow passed through the pipe. Figure 5 shows the copper concentration profile at the end of the stagnation period and before flushing (left panel) and the release curve of copper concentration as a function of the volume of water exiting the pipe during flushing (right panel). The continuous line is the result of the model and the Box-Whisker represents the measured field data. It can be observed that under laminar flow conditions, the release of copper decreases as the volume of water flowing through the pipe increases. When computing the mass of copper released, it is possible to confirm that this mass is larger than the mass estimated only using the concentration of copper in the bulk liquid within the pipe and its volume. The simulation was able to reproduce this behaviour, thus confirming the hypothesis that complexation of copper on the biomass is of importance when there is a biofilm covering the pipe wall. It also highlights the importance of hydrodynamics during flow conditions and stresses that diffusion transport is not the main mass transport mechanism.
Figure 5

Left panel: Dissolved and complexed copper in the pipe at the end of the stagnation period. Right panel: Release copper curve observed in the field (Box-Whisker) and computed with the model (solid line).

Figure 5

Left panel: Dissolved and complexed copper in the pipe at the end of the stagnation period. Right panel: Release copper curve observed in the field (Box-Whisker) and computed with the model (solid line).

One of the assumptions of the model is that the release of copper from the biomass during flow occurs instantly. There is not enough information regarding the kinetics of desorption and more studies would be required.

CONCLUSIONS

An advection-diffusion model can explain the copper release pattern only if a high copper concentration occurs near the pipe surface after stagnation. The results of the hydrodynamic model after the calibration process show that the best fit was achieved for an initial surface enriched concentration profile, since the diffusive profile does not agree with the observed concentrations of released copper. Therefore, the solid matrix on the inner surface serves as a storage compartment of labile copper that may be released under flow conditions, and behaves as a complex reactive surface (in the order of 100 μm thick) made up of biofilm material and abiotic precipitates. The biofilm matrix contains organic compounds (proteins, polysaccharides and humic substances) that are capable of binding copper ions during stagnation periods (Callot et al. 1987; Edwards et al. 2001; Beech & Sunner 2004). Thus, the diffusive transport from the pipe surface to the bulk during stagnation is not the only control of the flux of copper to the tap water when porous reactive microstructures cover the pipe.

Copper spatial distribution near the reactive layer, sorption and desorption mechanisms of copper on abiotic and biotic microscopic features, and detachment of nano-particles of corrosion by-products or biofilm from the surface (Vargas et al. 2010) might explain this behaviour. This highlights the need for models that consider the interaction between the hydrodynamics, chemistry, and structure at the solid-water interface to predict the release of corrosion by-products into drinking water.

ACKNOWLEDGEMENTS

This research was funded by FONDECYT projects 1110440/2011 and 1150357/2015. This paper was presented under the CONICYT-FONDAP project 1510020.

REFERENCES

REFERENCES
Beech
W. B.
Sunner
J.
2004
Biocorrosion: towards understanding interactions between biofilms and metals
.
Current Opinion in Biotechnology
15
,
181
186
.
Calle
G. R.
Vargas
I. T.
Alsina
M. A.
Pasten
P. A.
Pizarro
G. E.
2007
Enhanced copper release from pipes by alternating stagnation and flow events
.
Environmental Science & Technology
41
,
7430
7436
. DOI:10.1021/Es071079b.
Callot
P.
Jaegle
A.
Kalt
A.
Nanse
G.
1987
Pitting corrosion of copper tubes and carbon deposits
.
ESCS Studies Werkstoffe und Korrosion
29
,
519
522
.
Critchley
M. M.
Pasetto
R.
O'Halloran
R. J.
2004
Microbiological influences in ‘blue water’ copper corrosion
.
Journal of Applied Microbiology
97
,
590
597
.
Edwards
M.
Powers
K.
Hidmi
L.
Schock
M. R.
2001
The role of pipe ageing in copper corrosion by-product release
.
Water Science and Technology: Water Supply
1
,
25
32
.
Olivares
T. E.
Cienfuegos
R.
Vargas
I. T.
Pizarro
G. E.
2014
Experimental evidence for enhanced copper release from domestic copper plumbing under hydrodynamic control
.
Corrosion Science
80
,
473
481
. DOI:10.1016/J.Corsci.2013.11.064.
Pavissich
J. P.
Vargas
I. T.
Gonzalez
B.
Pasten
P. A.
Pizarro
G. E.
2010
Culture dependent and independent analyses of bacterial communities involved in copper plumbing corrosion
.
Journal of Applied Microbiology
109
,
771
782
. DOI:10.1111/J.1365-2672.2010.04704.X.
Pizarro
G. E.
Griffeath
D.
Noguera
D. R.
2001
Quantitative cellular automaton model for biofilms
.
Journal of Environmental Engineering, ASCE
127
,
782
789
.
Pizarro
G. E.
Garcia
C.
Moreno
R.
Sepulveda
M. E.
2004
Two-dimensional cellular automaton model for mixed-culture biofilm
.
Water Science and Technology
49
,
193
198
.
Pizarro
G. E.
Teixeira
J.
Sepulveda
M.
Noguera
D. R.
2005
Bitwise implementation of a two-dimensional cellular automata biofilm model
.
Journal of Computing in Civil Engineering
19
,
258
268
.
Pizarro
G. E.
Vargas
I. T.
Pasten
P. A.
Calle
G. R.
2014
Modeling MIC copper release from drinking water pipes
.
Bioelectrochemistry
97
,
23
33
. DOI:10.1016/J.Bioelechem.2013.12.004.
Schadler
S.
Burkhardt
C.
Kappler
A.
2008
Evaluation of electron microscopic sample preparation methods and imaging techniques for characterization of cell-mineral aggregates
.
Geomicrobiology Journal
25
,
228
239
. DOI: 10.1080/01490450802153462.
Vargas
I. T.
Pavissich
J. P.
Olivares
T. E.
Jeria
G. A.
Cienfuegos
R. A.
Pasten
P. A.
Pizarro
G. E.
2010
Increase of the concentration of dissolved copper in drinking water systems due to flow-induced nanoparticle release from surface corrosion by-products
.
Corrosion Science
52
,
3492
3503
. DOI:10.1016/J.Corsci.2010.06.027.
Werner
W.
1995
Uniform Copper Corrosion
.
Internal corrosion in water distribution system
,
Goteborg
,
Sweden
.