ICP Materials exposure sites have changed over the years. In this publication, only data from the following sites are considered. The selection was made from sites that are currently active and have more than just a few years of data. Selected sites include three industrial sites, nine urban sites and seven rural sites, in total nineteen sites:
Most of these sites were included from the beginning of the program (1987), except Katowice, Paris, Berlin, Svanvik and Chaumont, which were introduced later (1995–2000). It should be noted that these labels “Paris”, “Berlin”, etc. indicate that they are located in these cities. The values of corrosion, pollution and soiling at these sites should however not be considered representative of these cities, since the variation of corrosion and pollution within a city can be substantial.
Before presenting the results, a special note on exposure periods is needed. All exposures performed in ICP Materials so far have started in the fall, usually in October, and then lasted for one year or several years, also ending in the fall. In this paper, all exposure periods are labelled with the start year so that, for example, “1987” in reality is a short hand notation for a one-year exposure between the fall of 1987 to the fall of 1988, and “2011–2014” is a short hand notation for a four-year exposure between the fall of 2011 and the fall of 2015.
2.1. Trends in Air Pollution
The concentrations of the pollutants SO2
in air (µg m−3
), as measured at the ICP Materials sites, are included in the description of trends in air pollution. Figure 1
shows that the average of the SO2
concentrations at the industrial sites was considerably higher than for the urban and rural sites, in all the measurement years since 1987, except in 1995. In 1995, the value at the urban sites was much higher than in the measurement years just before and after, and nearly as high as for the industrial sites. The reason for this is technical and coincidental. It is due to a change of sites in the measurement programme, with only one urban site remaining in 1995, before a set of the former, and one new, urban sites were again included in 1996. This highlights a difficulty presenting averages and is the reason why this paper also presents trends for individual sites to illustrate characteristic trends. Until 1997, the average SO2
concentration measured at the industrial sites was above 30 µg/m3
. The average SO2
concentration measured at the urban sites was close to the average values measured for all the sites, in all years. By 1997, it had been reduced to below 10 µg/m3
, and, by 2005, to below 5 µg/m3
. The average SO2
concentration measured at the rural sites was below 5 µg/m3
in all the years.
shows a quite different situation for NO2
as compared to SO2
. The average NO2
concentration has not changed much since 1987. Some decreases in concentrations were measured until approximately 2000, with little change thereafter, except some possible increase at the industrial sites. In contradiction to the situation for SO2
, the average concentration of NO2
at the urban sites was higher than at the industrial sites in nearly all years, with the coincidental main exception being 1995, when only one urban site was included in the average. The higher average concentrations of NO2
at the urban sites may have been due to more emissions from traffic and domestic heating than at the industrial sites. The average concentration of NO2
at the rural sites was generally below 10 µg/m−3
, but with somewhat higher concentrations and more variation between years before 2000.
shows a quite different situation for O3
as compared to SO2
. The trends for the average concentration of O3
measured for the sites were slightly positive, and slightly more so from 1987 to about 2000 than from 2000 to 2014. The average concentration at the rural sites was always significantly higher than that at the urban and industrial sites. The difference between the average concentration at the urban and industrial sites was always small, with alternating ranking between them. In 1995, there was only one urban site in the calculated average, due to changing of sites in the programme, which explains the coincidental high “average” value in this year.
Looking at Figure 1
, Figure 2
and Figure 3
, there is a change in the trend of air pollution around 1997–1999 where a steep change is replaced by a more modest change (decrease for SO2
and increase for O3
). Therefore, a compact way of summarising the data from all sites is to present diagrams with averages for the three periods 1987–1989, 1997–1999 and 2011–2014. This will more or less summarise the whole trend for all sites as well as give the opportunity to compare changes for different periods (1987–1989 vs. 1997–1999, 1987–1989 vs. 2011–2014, and 1997–1999 vs. 2011–2014). Figure 4
, Figure 5
and Figure 6
show this type of diagram for SO2
shows that the decrease of SO2
at the industrial sites for the whole period of measurements, from 1987–1989 to 2011–2014, was dominated by a decrease in Kopisty, but from 1997–1999 by a decrease in Bottrop. The decrease at the urban sites from 1987–1988 to 2011–2014 was dominated by a decrease in Milan and Prague, with somewhat less decrease in Rome, Venice and Madrid. Since 1997–1999, the decrease was larger in Prague than in Milan and the ranking of the other sites also changed. The decreases at the new sites in Paris and Berlin were higher than in Madrid and Venice. The few measurement values for Rome showed a situation there after 1997–1999 was more similar to that of the low values in Stockholm and Oslo. It can however be noted that a relatively high SO2
concentration as compared to previous years, of 4.4 µg/m3
, was measured in Oslo in 2014. This points to the need, still, for measurements of SO2
in the present situation with generally low concentrations, and for attention to possible new or reappearing emission sources. Some notable decreases in SO2
since 1987–1989 were also observed at the rural sites.
shows a decrease in the NO2
concentrations measured at most of the sites since 1987–1989, with an overall correspondence to the decrease for SO2
, with the highest decrease in Milan, but with some noteworthy exceptions. Foremost among them, in Prague and Madrid, there was a small increase in the concentration measured for NO2
since 1987–1989 and a higher increase since 1997–1999. In both cases, NO2
was observed to decrease until 2000 and increase thereafter. During 1997–1999, NO2
was measured at four more sites, Paris, Berlin, Svanvik and Chaumont, than during 1987–1989. Measurements at the site Katowice started in 2000. In addition to the increase in NO2
in Prague and Madrid, the concentration measured for NO2
has also increased in Kopisty and Rome since 1997–1999. The reasons for these increases were high values for 2014 in Kopisty and 2011 in Rome.
The decreases in measured NO2 in Stockholm and Oslo since 1987–1989, and in Casaccia and Toledo since 1997–1999, as compared to the other sites, were relatively larger for NO2 than for SO2. Among the new sites included from 1997–1999 to 2011–2014, a large decrease in NO2 was measured in Paris.
Notable decrease in NO2 was measured for the rural site of Toledo since 1992, and since 1997–1999. A general trend of slightly decreasing NO2 was measured at the rural sites since 1987–1989 and 1997–1999, but with some high values for some years and variation in values for some sites, which could influence trend calculations. Most notably, at the site Casaccia, much higher values were measured in 1997 and 1998 than in other years, giving increase in concentration to 1997–1999 and decrease thereafter.
shows quite large increases in O3
concentrations at the Italian sites Casaccia, Venice, Milan and Rome, and Spanish sites Madrid and Toledo from 1987–1989 to 2011-2014. The increases were mainly due to low start values in 1987–1989, but also partly due to high values during 2011–2014 in Casaccia, Toledo and Madrid.
At the sites in Rome and Casaccia, lower values were measured in the mid-1990s, and then considerable increases were measured towards 2011–2014. Significant increases were also observed from 1997–1999 to 2011–2014 for some sites outside of Italy and Spain, for which O3 measurement results then had become available: Berlin, Paris and Bottrop. The changes in O3 showed no clear correlation with changes in NO2 and SO2. The largest variation and overall increase in O3 concentration was measured for the rural site Casaccia. Except for this site, the sites with the largest increases were all urban sites. At the other rural sites, and the urban sites Stockholm and Oslo, small increases or decreases were measured. Exceptions from this were the two Czech sites, the industrial site Kopisty and the urban site Prague, where large decreases were measured from 1997–1999 to 2011–2014.
In contrast to SO2, NO2 and O3, measurements of HNO3 and PM10 were not started in the programme until 2002. Furthermore, PM10 is not a mandatory parameter to measure. Therefore, the data for HNO3 and PM10 do not permit evaluation of long-term trends in the same manner as for SO2, NO2 and O3. Instead, data for individual sites are presented, in a similar way as for SO2, NO2 and O3, but using only two periods: 2002–2005 and 2011–2014. With measurement results for PM10 and HNO3 for only four to five years, any interpretation of trends should be made with caution.
shows the concentration of gaseous nitric acid (HNO3
). Comparing the two periods, the average concentration measured for most sites has decreased. There is considerable variation between the values for the industrial, urban and rural sites, with higher values measured at some rural sites than at the urban and industrial sites. The highest values were measured at the urban sites Paris, Milan, Venice, Rome and Madrid, and the industrial site Katowice. Notably, low values were measured at the rural sites Svanvik, Aspvreten and Birkenes, and at the urban sites Oslo, Stockholm and Berlin. Details on the measurements of nitric acid have been presented elsewhere [9
shows the concentration in air of particulate matter with aerodynamic diameter smaller than 10 µm (PM10
). It is not possible to make any general conclusions regarding trends based on this limited dataset. Among the urban sites, notably low values were measured at the sites of Stockholm and Madrid. Among the rural sites, the highest concentrations measured overall were in Toledo, and the lowest concentrations were measured in Birkenes and Svanvik. Some sites showed notable trends of decreasing PM10
, including the urban site Berlin and the rural sites Chaumont and Lahemaa.
In summary, the trends in the changes of the average concentration of SO2, NO2 and O3 for the ICP sites were all stronger in the first phase of the programme, from 1987 to about 2000, than thereafter. This was very apparent for SO2, clearly seen for NO2 for which there were no apparent trend after 2000, and slightly apparent for the positive trend for O3.
From about 2000 to 2014, a trend of gradual decrease in the SO2 concentration was measured at all sites, except Svanvik where a small increase was measured. The Svanvik site is located only 6 km away from the considerable SO2 source of the nickel plant in the town of Nikel, Russia. There was no general trend in the measured concentration of NO2 since 1997–1999, but for more of the sites decreases were measured than increases, with the largest changes in the measured concentrations being the decreases at the sites Milan (−40 µg/m3) and Paris (−33 µg/m3). Other sites where considerable decreases in NO2 were measured were Casaccia, Bottrop, Toledo and Stockholm. Considerable increases in NO2 were measured at the sites Prague, Madrid, Kopisty and Rome. A slight positive trend was measured for O3 from 1997–1999 to 2011–2014 for nearly all the urban sites, except Prague, and for the rural site Casaccia. Since 2002, a clear and continuous decreasing trend in the concentration of HNO3 was measured at all the industrial sites, at the urban sites Paris and Milan, and at the rural sites Casaccia and Toledo. Since 2002, a clear decreasing trend in PM10 was measured at the Berlin and Chaumont sites.
Taken together, for air pollution, the largest recent (since about 2000) decreases in concentration at ICP sites were measured in Milan and Paris, then Bottrop and then the other urban and industrial sites. However, considerable increases in NO2 were measured in Prague, Madrid and Kopisty, and considerable increases in O3 in Rome and Berlin. The changes for the rural sites were minor compared to the industrial and urban sites, but with notable decrease of NO2 in Toledo, and Casaccia, where the variation in NO2 between years and the increase in O3 was the largest.
2.2. Trends in Carbon Steel Corrosion
shows the average mass loss of unalloyed carbon steel (C < 0.2%, P < 0.07%, S < 0.05%, Cu < 0.07%) for one-year exposures for industrial test sites with one selected individual site, Kopisty. The corrosion decreased significantly between 1987 and 1997. The corrosion then remained on a level around 240 g/m2
for the first exposure year, corresponding to a 50% decrease compared to the original value. In the last 25 years, the pollution at Kopisty reduced tremendously because of the decline of heavy industry. ISO 9223 corrosivity category C3 ranges from 200 to 400 g/m2
. With a recent value of 210 g/m2
for the first year of exposure, this industrial test site has changed its corrosivity category during 1987–2014 from C4 to low C3.
In Figure 10
, the trend of carbon steel corrosion is shown for the nine urban test sites with Prague as a typical example. As for the industrial sites, there is a strong decrease of mass loss between the exposures in 1987 and 1997, but also a small reduction of mass loss in the following periods.
The trend for rural test sites is similar to the trends of industrial and urban test sites. For rural test sites, a slight decrease of mass loss values was found (see Figure 11
). Today, the urban test sites (C4 and C3, according to ISO 9223) are tending towards C2 with values typical for former rural atmospheres. This shows that labels such as “industrial”, “urban”, “rural” and “marine” can be useful for indicating the type of pollution (dominated by SO2
and chloride) but that they are not at all useful for classifying levels of corrosivity in a quantitative way.
Two repeated four-year exposures were performed, during 1997–2000 and 2011–2014. As for the one-year exposures, all test sites showed a decrease of corrosion values.
As an exposure was not performed during 1987–1990, a comparison of these two exposures does not show such a large decrease in corrosion rates, since the period of maximal reduction of air pollution (1987–1997) is not included. Figure 12
shows the mass loss after four years of exposure vs. mass loss after one year of exposure for two different exposure periods, 1997–2000 and 2011–2014. The clear relationship between one- and four-year periods shows that it is equivalent to show trends in corrosion based on four-year exposures instead of one-year exposures. The advantage of using four-year values for showing trends in corrosion is that it is less sensitive to year-to-year variation in climatic parameters (temperature, relative humidity, and precipitation). The higher corrosion values also make it easier to identify significant trends for individual sites. Note that in Figure 12
, the relationship between one- and four-year values is practically 1:2, corresponding to a square-root kinetics for carbon steel during the first four years of exposure.
shows trends of four-year corrosion at the test sites and that corrosion has decreased at all sites corresponding to average levels for 2011–2014 of about 60% of those for 1997–2000.
2.4. Trends in Zinc Corrosion
Zinc (>98.5%) with two different kinds of surface preparation, ground and glass blasted, has been exposed in the programme; the ground from the beginning of the programme, from 1987, which was then replaced with the blasted from 1997. Simultaneous exposures were performed in 2000 and 2008. The glass blasted zinc has a rougher surface leading to, at least initially, higher corrosion loss values.
As can be seen in Figure 16
, the mass loss after one year for the investigated industrial test sites decreased significantly in the period 1987–1997. From 1997, the value then remained at a constant level (around 10 g/m2
) for the first exposure year (blasted zinc). The mass loss values for Kopisty are shown as an example for an industrial site, where the ISO 9223 corrosivity category changed from C4 to C3 (ISO 9223 corrosivity category C3 ranges from 5 to 15 g/m2
In Figure 17
, the trend is shown for the nine investigated urban test sites with Prague as a typical example. As for the industrial sites, there is a strong decrease of mass loss between exposure periods 1987–1988 and 1997–1998, but also a small reduction of mass loss in the following periods. Today, the corrosivity category for the urban test sites (C3) is starting to approach C2 with values comparable to those in rural atmospheres.
The trend for rural test sites is slightly less distinct compared to industrial and urban test sites. There is a higher fluctuation of mass loss values from year to year. For some rural test sites, a slight decrease of mass loss values can be found but there are other sites, such as Lahemaa (Estonia), with no clear trend (Figure 18
). Higher mass loss values were sometimes measured at rural sites compared to urban sites. Overall, the mass loss values (blasted) at most sites now range 5–10 g/m2
. These values correspond to a thickness reduction of 0.7–1.4 µm, calculated based on the density of zinc 7.14 g cm−3
(thickness reduction in µm = mass loss in g m−2
Three repeated four-year exposures, starting from 1987, were undertaken with zinc samples in unsheltered exposure at different test sites. The first exposure period (1987–1988) was with ground surface condition and the other two with blasted surface condition. Figure 19
shows a comparison between four- and one-year corrosion, similar to for carbon steel (Figure 12
) and weathering steel (Figure 14
). In contrast to the other materials, there is a systematic difference when looking at the different periods. For ground zinc, the relationship between the mass loss of the one- and four-year samples is about 1:4, corresponding to a line going through the point 15 g m−2
(one year of exposure) and 60 g m−2
(four years of exposure), except for two sites. If the relationship between one- and four-year data were exactly 1:4, this would correspond to a linear development of corrosion with time, i.e., linear kinetics. For blasted zinc, the four-year values are lower than what would be expected from linear kinetics, indicating that the high corrosion values resulting from blasting as opposed to ground is an initial phenomenon most prominent after one year of exposure.
shows all four-year data for the individual sites. The first value (1987–1990) is not directly comparable to the two later values due to the different surface treatment. However, it is not possible to make a correction factor since, parallel four-year exposures have not been performed. Furthermore, results presented in Figure 19
show that it is not possible to use a comparison of one-year values to derive a correction value for four-year samples. Nevertheless, it has some merit to present ground zinc (uncorrected) in the same diagram as blasted zinc, even if the decrease in corrosion between 1987 and 1990 and the other periods will be underestimated in this way.
An example of a strong decreasing trend is the test site Kopisty. The mass loss at the first exposure period (1987–1990) was around 60 g/m2
during four years (which means an average yearly corrosion rate of 15 g/(m2
year) or 2.1 µm/year). This value decreased to 23 g/m2
for the third exposure period (2011–2014). The corrosion rate at this test site now lies at 5.8 g/(m2
year) or 0.8 µm/a and can now be characterised as C3. This means a reduction of the corrosion rate of 66% from the value of the first exposure period. In practical terms, this means that a galvanized steel structure with a typical zinc layer thickness of 80 µm would previously have shown red rust (1987) after approximately 30 years, while it would now take 100 years. It should be noted that it is difficult to accurately estimate the long-term corrosion rate based on these values, especially considering the results presented in Figure 19
, but, in any case the reduction is substantial. Furthermore, values at urban test sites are now in the same order of magnitude as rural test sites, which makes it more difficult to evaluate the effect of air pollution based on zinc corrosion.
When looking at Figure 20
and comparing the two last periods (1997–2000 and 2011–2014), there is a decrease in corrosion at all sites where a comparison is possible but it was not possible to capture this overall trend based on one-year data only (Figure 16
, Figure 17
and Figure 18
). Thus, the advantage of using four-year values for showing trends in zinc corrosion is quite evident. Four-year values are less sensitive to year-to-year variation in climatic parameters. The higher corrosion values also make it easier to identify significant trends for individual sites.