CASE STUDY ON COPPER CORROSION

ABSTRACT Corrosion of copper plumbing tubes indicated by either the formation of 'blue water' or pitting corrosion, is a phenomenon which randomly affects large parts of the east coast of Australia, as well as New Zealand, USA and Europe. It is characterised by the production of voluminous blue/green corrosion products in the tube bores and is primarily associated with cold, soft unbuffered waters of high pH and negligible levels of disinfectant residual. The history and causes for this type of corrosion are described, with some emphasis placed on the various erroneous statements that have been advanced to explain the problem. The corrosion mechanisms whilst by no means well understood are shown to be associated primarily with water chemistry factors together with a contribution from copper tube surface condition. Management of the corrosion problem is shown to be achievable, using two case studies recently completed from regional NSW. Key words: measures. Theoretical Corrosion can only take place if certain elements are present; iron, oxygen and water, in the form of Moisture in the air. If a section is hermetically sealed the moisture in the entrapped air will allow only a limited amount of corrosion to take place. As the oxygen is used up oxidation will cease, since the entrapped air cannot be replenished. Copper tube, corrosion, water quality, remedial

Practical. On the basis of this theory it seems obvious that the protection of hollow sections is simply a question of air tightness. Thus the general outlines of the investigation are clearly defined; the inspection of structures which have been standing for many years must prove that the normal methods of fabrication are sufficient to prevent internal corrosion. This report investigates the problems presented by the welding of hollow sections used in structures and in particular the protection given to the interior of sections when closed by welding. On the basis of his own observations and on foreign experience, he reaches the conclusions which we reproduce in full below. 1) Sealed hollow sections require no internal protective coating and may be regarded as essentially immune from corrosive attack. 2) Condensation in a sealed section is impossible, and when found upon inspection is evidence of an opening having developed -possibly a small opening that is drawing surface water in through capillary action. 3) Adding a "pressure equilibrium hole" at any point in a hollow structure where water cannot enter by gravity will prevent aspiration in an imperfectly sealed system. (If the engineer has qualms about condensation, he might as well put the hole at a low point where it would serve for drainage also - merely to satisfy his peace of mind.') 4) An "open" system should generally be kept as tight as feasible with rubber gaskets used at manholes and such closures positioned so as to avoid water accumulation and the possibility of its entrance by capillarity action or aspiration. A strategically placed pressure equilibrium hole might be advisable. Such Systems should be protected with an interior coating.

5) A ventilated hollow structure should be internally protected and have adequate ventilation holes at each end and in the sides. 6) Bolt and rivet holes should be avoided where-ever possible; they create conditions conduciveb to water entrance by capillarity action. In general, good design and good practice should eliminate concern about corrosion in hollow steel sections. The overwhelming mass of evidence, scientific analysis, and European experience suggest that it is - as the Englishman said - more of a "bogey" than a serious engineering problem.

INTRODUCTION Background to Copper Corrosion L & T has performed fabrication surveys on 10 copper corrosion, at three different fabrication yards. In the interest of protecting our client, names of the installations or fabrication yards are not revealed in this paper. Typically, fabrication surveys would include the following to identify future (when operational) integrity and inspect ability issues Although copper tubes for carrying potable water are known to have existed 4,500 years ago (Nicholas, 1994), it is only with the post-WWII introduction of thin-walled capillary tubing that copper has become the default plumbing material used in virtually all first-world countries, including Australia. Figure 1:Membrane Theory (Lucey, 1967)

Whilst this theory was originally generated for relatively cold hard waters on tubes often associated with carbon surface films, called Type 1 pitting, Lucey himself (Lucey,1982) suggested a commonality with all types of pitting, including hot waters and cold, soft waters. However, aspects of Lucey's membrane theory have been questioned (Edwards et al, 1994) and the mechanism of both nucleation and growth of pits are now unresolved. Excellent reviews of pitting corrosion of copper tubes have been published (Edwards et al, 1994, Sequeria 1995) but although the quantity of published literature is voluminous the precise mechanisms are still not well understood. However, work on pitting corrosion continues, with a significant component of this being carried out in Australia and New Zealand. These corrosion events, themselves often associated with pitting failures, are observed in soft, cold unbuffered water supplies that predominate along the east coast of Australia. The NSW Land and Water Conservation report itself detailed the results of a survey across the whole of NSW and, given a 52% response rate, the results showed that about 30% of Councils that responded recognised a significant problem.

A central question in all of these studies is why, after so much research, are the prime causes and solutions to both 'blue water' and pitting corrosion poorly understood? Primarily, this is due to both the complexity of the phenomenon and the large number of variables that may be involved. Another principal factor is the randomness of corrosion: there is rarely more than a few percent of buildings in a particular location involved, and it is often found that failures have a tendency to cluster in certain suburbs or even streets. The situation is rendered even more complex by the tendency for interested parties in the corrosion issues - often but by no means restricted to the actual building owners - to attempt to identify spurious and almost invariably misleading causes for copper corrosion. These are usually single issue variables which often attempt to both discredit the local water authority and apportion blame for the incidents. In fact, the issues are quite complex enough and the following listing may assist in identifying some of these spurious causes that are often erroneously quoted as causing copper corrosion. Non-contributory issues Stray Electrical Currents - Stray currents do exist and can readily be measured on copper tubes as they are usually connected to the earth of the Electricity M.E.N. distribution network. Unfortunately, these currents have nothing to do with internal corrosion as demonstrated by a number of studies (Moss and Potter, 1984). It is still regularly quoted as a source of corrosion.

Sources of Water - Recently, there was community concern at one corrosion site that 'blue water' was caused solely by the use of bore water. In fact, both surface and underground sources are involved with copper corrosion. Fluoride - Shown in a 1980 report to have absolutely no influence on corrosion (Bensley, 1980). Grade of Copper - All copper tube has been manufactured from phosphorus deoxidised copper since before WWII. In itself it has no influence, but the surface condition of the tubes is another matter and has a significant role.

BASIC CRITERIA FOR COPPER CORROSION All potential explanations of copper corrosion, either pitting or 'blue water', must encompass the following observations: Failure is almost never found in relatively hard water areas such as Adelaide and Brisbane. It is confined to soft, poorly buffered water supplies.

'Blue Water', and the majority of pitting cases, are only found in cold water.

Tube surface conditions have clearly been shown to have an effect on copper corrosion, with some temper conditions worse than others (Taylor et al, 1998). However, corrosion has been frequently identified with all forms of tubing and surface condition (Page, 1973; Nicholas, 1980).

 It is very random in occurrence even in areas where water characteristics and tube surface condition might predicate copper corrosion. In terms of incidents per 100 properties it rarely exceeds 5%. In any given distribution system, copper corrosion incidents are more likely to occur at the extremities of the network. Several factors have been put forward as direct causal explanations of these observations. These can roughly be divided into water chemistry factors and tube surface condition factors: Condition of Tube The original work on copper pitting in the UK firmly identified so-called 'carbon films' as the basis for tube pitting (Campbell, 1950). These were glassy, black coherent films of carbon produced by a furnace anneal prior to the final drawing process so called 'half hard' or bendable quality (BQ) tubes. The remedy was to remove the films using abrasive cleaning.

During the 'blue water' episodes investigated in Australasia in the 1970's (Page, 1973., Nicholas, 1980) it was clear that the mainly hard-drawn tubes involved in these corrosion incidents did not have these carbon films, and the influence of tube surface became less of an issue. Nicholas, (1980) claimed that carbon films were present to a degree on all tube surfaces and a causal relationship between the presence of carbon and corrosion could not be established. More recently, extensive laboratory trials at the CSIRO have indicated that tubes of different temper do give different results in simulated 'blue water' tests (Taylor et al, 1998) although the precise reasons are not at all clear and continue to be under investigation. It can be said, however, that tubes with a thermal history (i.e. annealed or BQ) show a greater tendency to corrode when compared with hard drawn tubes, at least in the controlled testing carried out by the CSIRO.

Water Chemistry Regardless of the possible influence of tube surfaces in copper corrosion, in Australia it has been found that for corrosion to occur, the prevailing water chemistry must have certain characteristics. Tubes, of whatever surface condition, simply do not cause 'blue water' or pitting to perforation in either Adelaide, Brisbane or wherever the source water is moderately hard. Cuprosolvency or minor surface pitting can be found in these waters, but rarely is this a serious corrosion problem. The extent of copper corrosion on the east coast of Australia, Perth, Tasmania, New Zealand and elsewhere encompasses a range of 'soft' water chemistries. All areas of failure, however, seem to share significant characteristics:

 Soft, with Total dissolved solids <300mg/L Low alkalinity (typically, <30mg/L) Ability to leach lime from cement lined cast iron pipes leading to relatively high pH at the end of the distribution system. Almost invariably cold water, or <50c if warm water involved. Varying levels of chloride and sulfate, but corrosion is known to occur with both these analytes <7mg/L. Low residual disinfectant. Anecdotal arguments can be advanced, largely from these simple observations coupled with the randomness and previous factors discussed in section 1.3., that the prime cause of 'blue water' is Microbiologically Induced Corrosion (MIC). This mechanism postulates the colonisation of newly installed tubes by various strains of copper tolerant bacteria (Wells, 1999) which themselves produce acidic interfacial layers on the tube surface, thus lowering surface pH and altering the protective qualities of the oxide film. Whilst this mechanism is by no means universally accepted it has been shown (Taylor et al, 1998; Wells, 1999) that chlorine or chloramine disinfectant is highly successful in controlling 'blue water'. Nevertheless, alternative chemical explanations for the beneficial effects of disinfectant residuals are possible, if currently not well-explained, and definitive experiments are planned in the near future. Further, the MIC theories currently only exist for 'blue water' events, not for pitting corrosion. Given the clear links between the two types of corrosion, it would be unusual if some commonality of cause did not exist, but this is unproven and also under active investigation.

Case Study I Shoalhaven Council has experienced significant pitting corrosion in a relatively isolated community at Sussex Inlet. Investigation showed that the soft, unbuffered water supplying Sussex Inlet was at the end of a long cement mortar lined distribution system with subsequent long retention times. Consequently, water supplied was at elevated pH and negligible disinfection residual. Relatively high levels of pitting corrosion were reported and most significantly, this was not reported from population centres upstream with ostensibly the same copper tubing and the same basic water chemistry. Detailed surveys of the Sussex Inlet showed a total tube failure rate of about 5%-7%, which included cases of 'blue water' previously unreported. Experience here and elsewhere shows that actual failures or incidents are usually many times higher than those reported to the relevant utility. The outcome here is the introduction of full-scale calcium bicarbonate dosing of the bulk supply together with the maintenance of effective disinfectant residual. Case Study II Rouse Water, a bulk water supplier to a number of Northern NSW Councils, received complaints of 'blue water' from the Evans Head community in May, 1998. As with case Study I, the water caused few if any failures at the major population centre (Lismore) but significant incidents when the same supply finally reached downstream at Evans Head. The 'blue water' complaints were relatively numerous, but still tended to cluster at key streets within Evans Head itself. Large parts of the community were unaffected.

A very comprehensive corrosion management plan has been implemented by Rous Water, perhaps the most significant field trials yet attempted to alleviate 'blue water'. It can best be summarised through Figure 3 below: One of the early attempts to control corrosion on an individual premises basis was to carry out hot water flushing. This was highly successful in immediately lowering copper levels, but gradually these returned in the majority of cases to the previous high levels. Whilst a chemical theory explaining these observations is possible, the alternative MIC explanation of a bacteria kill-off from the hot water treatment followed by gradualre colonisation and thus increases in copper levels is arguably a more attractive theory. Whilst carbon dioxide dosing on its own successfully dropped pH levels in the distribution system from around 9.5 to 8.0, Figure 3 clearly shows that there was no concurrent or consistent drop in copper levels. The combined use of line/carbon dioxide, however, shows a relatively immediate drop in average copper levels, indicating that controlling pH alone does not necessarily control copper corrosion. The added buffering of lime/CO2 may control corrosion by simply making localised changes in pH on the copper surface (from whatever mechanism) harder to achieve. Finally, the re chloramination introduced in April, 1999 achieved another significant drop in copper such that average levels are now below 0.2 mg/L, an order of magnitude less than the recommended guideline limit of 2mg/L. Figure 3 shows that it took over a month for the secondary disinfection to significantly lower copper levels. This is due to the deliberately low levels (<0.1 mg/L) achieved at the customer tap so as to minimise taste and odour complaints. Nevertheless, 'blue water' has been controlled.

Whilst the beneficial effects of these management processes on the 'blue water' phenomena can readily be assessed through chemical analysis of the 'first flush' water samples, this does not necessarily indicate that corrosion processes are not continuing on the tube surface in the form of pitting corrosion and/or build-up of blue/green corrosion product on the surface, both of which were known to occur at Evans Head. To investigate these possibilities, a number of 19mm OD tube samples, of different tempers and various experimental surface treatments, were installed in 2 metre lengths in series at a site in Evans Head that had previously experienced pitting corrosion. The tubes were installed about a month before lime/carbon dioxide dosing came on line (see Figure 3) and thus had exposure to undoes water for this initial period. 300mm lengths of each tube were removed after 1 month, 3 months and 6 months exposure respectively for sectioning and subsequent examination. Whilst this examination is so far incomplete, some general observations can be reported: Some tube samples, in broad confirmation of the previous CSIRO trials, (Taylor et al, 1998) are far more susceptible to corrosion than others.

Many of the susceptible tubes were showing clear signs of surface corrosion after only one (1) month exposure in undoes water.

The onset of bicarbonate dosing was not successful in preventing the growth of pits, or the formation of corrosion product, on susceptible tubes after a further two months of exposure. It is not clear-cut that tubes with higher levels of carbonaceous films automatically suffer most from corrosion.

When comparing the appearance of tube surfaces after three and six months exposure, there are indications that the corrosion rate has generally been slowed down by a significant margin.This trial is now being extended to assess the longer term performance of these tubes. The samples previously removed are currently being subjected to a range of physical and chemical surface analysis techniques in order to reach some quantitative conclusions as to the basic reasons for the differing performance.

Further, in order to assess the performance of these susceptible tubes when exposed in their unused state to calcium bicarbonate dosed and rechloraminated water, a second series of tubes have recently been installed at the same site. This second round of trials also have a parallel loop in which a duplicate set of new tubes have been fitted which are connected to an activated carbon filter to remove the disinfectant. In this way the effects of secondary disinfection can be assessed. CONCLUSIONS Corrosion of copper plumbing tubes can occur in an unpredictable manner whenever soft waters predominate. It is encouraged when the supplies are at the extremities of distribution systems where relatively high pH and low residual disinfection is found. The mechanisms of 'blue water' and pitting corrosion are not well understood, although an MIC mechanism can be tentatively argued, at least for 'blue water'. The tube surface has a secondary but important role to play in the corrosion process, although the precise factors are also not understood.

 Regardless of the mechanisms involved, 'blue water' at least can be controlled by a corrosion management program that provides buffering of the supply on either a localised or global basis together with maintenance of an adequate disinfectant residual. REFERENCES Andrews, C (1996) ' Copper pipe failure and blue water survey' Uni of Newcastle Department Chemical Engineering, March. Bensley, B. (1980), 'Fluoridation and copper corrosion' Hunter District Water Board, internal report. Campbell, H.S.(1954) Proc. Soc of Water Treatment and Examination, 8, 100 Campbell, H.S (1950) 'Pitting corrosion in copper water pipes caused by films of carbonaceous material produced during manufacture'. Jnl.Inst.Metals, 77, 34.5. Edwards, M., Ferguson, J.L., and Reiber, S.H. (1994) 'On pitting corrosion of copper', Jnl. AWWA, 86, 74. Lucey, V.F. (1967) 'Mechanism of pitting corrosion of copper in supply waters'. Br.Cor.J., 2, 175-185. Lucey, V (1982) 'Assessment of type 1 pitting corrosion characteristics of potable waters' Corrosion of Copper and Copper Alloys in Building Symposium, Jpn CDA, Tokyo. Mattsson, E. and Fredriksson, A.M. (1968) 'Pitting corrosion in copper tubes-cause of corrosion and counter measures' Br. Corros.J, 3, 246-257. Moss, G., and Potter E.C. (1984) ' Interactions between cold potable water and copper tubes' CSIRO Concluding report No. 1534R.

Nicholas, DMF (1994) Arsenic and old brass' Australasian Corrosion Association Conference 34, Adelaide, South Australia. Nicholas, D.M.F. (1980) ACA Conference 20, Adelaide, November. Nicholas, D.M.F. (1980) ' Analysis of the surface condition of copper tubes' Jnl HDWB, Autumn. Nicholas, DMF (1987) 'Corrosion control in Hunter waters; effects of calcium bicarbonate dosing' HWB internal report, Newcastle, NSW. NSW Department of Land and Water Conservation (1999) ' Copper corrosion in country NSW Survey and Management Strategies' Internal Report, Urban water cycle planning unit, May. Page, G.G. (1973) New Zealand Journal of Science, 16, 329 Sequeria, C.A.C. (1995) 'Inorganic, physiochemical and microbial aspects of copper corrosion': literature survey. Br.Cor.J., 30, 137-153. Taylor, R.J., O'Halloran, R.J. Sexton, B.A. and Smith, F.L. (1998) 'Blue-green water investigations undertaken by CSIRO for City West Water'. CSIRO report CMST -C-C-98-21, August. Wells, B. (1999) ' Review of remedial treatments for Blue Water, pitting and cuprosolvency' WSAA restricted report.