Professional Documents
Culture Documents
Life Cycle Predictions For Coating Systems
Life Cycle Predictions For Coating Systems
ABSTRACT
In-service coating performance data is often collected in an effort to predict coating life
expectancy and time to failure. Having a general knowledge of coating life expectancy
has many benefits, such as being able to project future maintenance requirements, pri-
oritize repairs, estimate the impact of application parameters on the coating life, and
quantify the risk of deferred maintenance. The DoD maintenance communities benefit
from knowing which assets are a priority for repair as well as the performance factors
influencing coating life. Such data should also be used to asses the design of new
equipment.
The following paper addresses how to use engineering data to illustrate a coating life
cycle and subsequent materiel degradation. Based on the user’s definition of failure for
the asset at hand, a coating life can be defined. This process includes looking at mul-
tiple factors that influence the degradation of coatings and the asset.
KEY WORDS
Life cycle, prediction, coating life expectancy, coating failure, color change, substrate
corrosion, time-to-failure, incubation time
INTRODUCTION
One of the questions often asked / considered when selecting a coating system for an
asset is, “How long is this coating going to last?” Unfortunately, this is a very difficult
question to answer and the response will often vary depending on the source. For in-
stance, a paint salesman and a coatings chemist may have drastically different an-
swers, as they each have a different perspective of the situation. The problem is that
neither of these responses may be applicable to the current situation. The only way to
get a suitable answer to this question is to be honest about the coating expectations
and define what constitutes a coating failure.
There are two important concepts in defining coating failure. First, the definition should
be asset-specific. Take an aircraft and a trailer for example. The coating “failure crite-
ria” for a trailer may have a greater allowance for degradation than that for an aircraft
because trailers are routinely scuffed and scratched in their normal use. On the other
hand, an aircraft will have little degradation allowance as its coating failure criteria due
to various concerns including airframe corrosion. If the aircraft criteria were applied to
the trailer, it would result in unnecessarily increased labor and maintenance costs.
However, if the trailer criteria were applied to the aircraft, the aircraft could lose functio-
nality. Therefore, it is important to consider the specific asset at hand when defining
coating failure. The second concept is that the failure definition should specify an objec-
tive performance criterion. Choose performance criteria which is most important to the
situation and tailor the definition accordingly. Some examples of performance criteria
1
include color/gloss retention, coating adhesion, cutback and structural corrosion subse-
quent to coating deterioration. Once a performance criteria is chosen, set a limit as to
what is acceptable. Only after establishing performance criteria and a failure point can
one move forward to predicting the coating life expectancy.
Once a performance criterion and a failure point are set, predicting the coating life for a
specific asset can be derived through a review of coating and corrosion data applicable
to the coating, equipment, and exposure environment (service) combinations of con-
cern. Simple statistical techniques will allow for the analysis of this data and prediction
coating performance. This will aid in predicting what percentage of an asset is going to
fail at a particular time, how often the asset is going to need maintenance, and the type
of corrosion maintenance that the system is likely to require. All of these will aid in pre-
serving and sustaining our assets to their full life potentials.
The first step is to establish the behavior of the coating system over time concerning the
performance criteria of interest. Depending on the performance criteria, there may be
several phenomena to illustrate in the coating performance timeline. Consider an asset
where the failure point is defined by a given amount of structural metal loss and that this
asset is coated for protection. The first phenomena or performance criteria to illustrate
is the time for the formation of initial coating holidays: an “incubation period.” The
second performance criteria would be substrate corrosion following the formation of
these holidays. Combining data for these two phenomena gives a full picture on the life
cycle of a coating. It should also provide a time estimate for repairing the coating to
avoid the risk of substantial structural damage.
The following three case-studies illustrate how these life cycles may be developed.
2
green (+a to –a), and yellow to blue (+b to –b). In reality a ΔE of any value can be a
combination of a change in one or more of these parameters.
It can be seen that as the ΔE changes, the color varies greatly. Too much of a color
shift could render the paint useless. In order to avoid large color shifts, some specifica-
tions define a failure point as a ΔE of + 2.0. Using a ΔE of + 2.0 as the failure point, it is
possible to seek out applicable in-service performance data for color change and de-
termine a time to coating failure.
In one published example of coating color change1 data were developed for color
change versus time for four MIL-Spec coatings exposed in Arizona and Florida for up to
97 weeks. Figure 2 presents the change in color over the exposure period for the four
coating systems averaged for the two exposure sites.
3
Figure 3 uses this same dataset and shows a plot of the time to a color change equal to
a ΔE of 2.0 for each of the 4 coatings in each of two locations—eight data points. As
shown in the plot the data range from 0.4 years to 3 years for nominally 10% to 90% of
the population.
P er cent
M edian 1.27399
P DF
IQ R 1.27572
10 F ailure 8
0.15 C ensor 0
A D* 1.607
C orrelation 0.979
0.00 1
0.0 1.5 3.0 4.5 0.1 1.0 10.0
Y ear s to Failur e Y ear s to Failur e
1.0
Rate
50
0.5
0 0.0
0.0 1.5 3.0 4.5 0.0 1.5 3.0 4.5
Y ear s to Failur e Y ear s to Failur e
The data were subject to a least squares regression against a Weibull distribution func-
tion. The pertinent statistics for this analysis are displayed in the figure. From the sur-
vival function plot in Figure 3, one can see that with our failure point of a ΔE of 2.0, 50%
of these polyurethanes will fail within ~ 1.5 years. These distributions allow for the
compilation of a time to failure for any given percentage of exposures for this coating
system.
Using this information, the maintenance planner can identify as a function of time what
percentage of their fleet will require repainting for color / camouflage reasons. From
this, planners can then nominate assets for painting based on priorities, such as active
duty requirements or to make the most efficient use of their facilities (e.g., if using a mo-
bile repaint service for reserve units, this data suggests scheduling painting every 2-3
years minimizes mobilization costs per asset).
4
ballast tank which is coated to mitigate the structural loss of steel. One performance
criterion is the breakdown of the coating. The other is the corrosion of the underlying
substrate. The failure point cannot be defined as the point in time for the first occur-
rence of coating breakdown—the first area of rusting. This may occur shortly after
launch. The real concern might be predicting the cumulative occurrence of coating fail-
ure followed by substrate corrosion.
IQ R 4.05047
10 A D* 1.839
0.05
C orrelation 0.923
0.00 1
0 5 10 15 1 10
Incubation T ime Incubation T ime
0.4
P er cent
Rate
50
0.2
0
0.0
0 5 10 15 0 5 10 15
Incubation T ime Incubation T ime
5
This analysis of coating breakdown shows that after 5 years in service more than half of
the coatings will have begun to deteriorate. It also shows that the coatings are likely to
begin to deteriorate at about 4 years, yet a small fraction may last beyond 10 years be-
fore starting to deteriorate. The parameters for the Weibull distribution (the Shape and
Scale factors) dictate all of the parameters of these data plots.
For our current purposes, we choose to simplify the Melchers model and simply focus
on the short term and longer term steady state corrosion rates. The initial corrosion
rates (for times ~ < 3 years), are approximated to be from 0.09 to 0.31 mm/yr. Melchers
reports the longer term corrosion rate to range from 0.05 to 0.125 mm/yr. In knowing
these values, it is possible to start to put together a timeline for metal loss corrosion for
times less than and greater than 3 years.
To reduce the complexity of Melchers data and modeling, it is possible to use two cor-
rosion rates—(1) an initial corrosion rate for the first three years of exposure and (2) a
corrosion rate beyond three years. Using Melchers’ data from a variety of seawater ex-
posure conditions, Figure 5 was constructed to show histograms of these corrosion da-
ta.
6
Initial Corrosion Rates < ~ 3 years Corrosion Rates > ~ 3 Years
Melchers (Ro) Melchers (Rs)
30
Mean 0.0733
20
Mean 0.1846 StDev 0.02455
StDev 0.06646 N 10
25
N 14
15
20
Percent
Percent
15
10
10
5
5
0 0
0.05 0.10 0.15 0.20 0.25 0.30 0.02 0.04 0.06 0.08 0.10 0.12
r01 rs1
Figure 5: Histograms of Corrosion Rate Data, Steel in Seawater before and after
3 Years
Assuming that these corrosion rates begin following the initial breakdown of the coating,
combine three elements in a Monte Carlo model4. These elements are the time to coat-
ing breakdown (Figure 4) and the probable corrosion rates (Figure 5). In the Monte
Carlo model, a random number generator function was used to select a time to coating
breakdown following the equations predicted by Figure 4. Using a second random
number, a corrosion rate for the first three years after the first coating breakdown was
selected from the histogram in Figure 5 (left chart); for times extending beyond 3 years,
a second corrosion rate was determined from Figure 5 (right chart). The author as-
sumes a total service time of 15 years. The total corrosion time is thus 15 years less
the incubation time. The total corrosion is the corrosion rate for the time period times
the exposure time in the < 3 year region and the > 3 year region. Obviously for expo-
sures less than 3 years (beyond coating breakdown), there is only one contribution to
the model. Figure 6 shows the results of this model.
7
Predicted General Corrosion Rate - 15 Years Ballast Tank
16
14
12
10
Percent
0
0 1 2 3
Total Corrosion - mm (1 mm = 40 mils)
Figure 6: Predicted Total Corrosion Rate for Seawater Ballast Tank at 15 Years
Figure 7 shows the data obtained and plotted in this paper (as the author’s Figure 2).
As is noted, there is a considerable spread in the data sets. The data for the 15 year
data are of particular interest.
8
Figure 7: Corrosion Depth vs. Ship Age – Seawater Ballast Tanks
Figure 8 is the 15 year graph from the upper left quadrant of Figure 7 compared to the
values predicted by the above analysis (Figure 6).
14
12
10
Percent
0
0 1 2 3
Total Corrosion - mm (1 mm = 40 mils)
While the agreement is not perfect, the ranges of expected corrosion are similar. Per-
haps the predicted cumulative corrosion damage (for Navy ships) is somewhat more
than is experienced on the commercial side, though commercial ballast tanks may op-
erate at different filling cycles than the ballast tanks on Navy. Of course there are other
potential differences—the influences of which can be both positive and negative.
9
Pragmatically, the analysis seems to put a maintenance engineer into the correct range
of expected corrosion.
If the failure point for the ballast tanks is substantially less than corrosion of 2-3 mm
(maximum), then the coatings need to be replaced at an interval much less than 15
years.
The pertinent data of interest would be the corrosion rate of steel within this crevice
joint. From work performed as part of an ongoing study5,6, the corrosion occurring with-
in uncoated crevices exposed to marine environments was studied. This study showed
that the nominal corrosion rates for steel in such tight crevices ranged from ~ 2 to 14
mpy in one year, from 5 to 20+ mpy in two years, and generally exceeded 20 mpy after
three years. The maximum corrosion rate that could be measured was 20 mpy. Figure
10 provides a picture of the test samples after three year exposure. The figure on the
left is the steel test coupon; the figure on the right shows the initial configuration.
10
Figure 10: Crevice Corrosion Samples after 3-Years Marine Exposure
This data can be compared to the experience of the USMC with trailers in the harsh ma-
rine environment of Okinawa, Japan. The USMC Corrosion Prevention and Control
Program Office maintains an extensive database of inspection data on the variety of
ground vehicles and associated tactical assets fielded at different Marine Expeditionary
Force sites. One of the corrosion phenomena that are tracked in this database is the
presence of crevice corrosion. The severity of the crevice attack observed is rated on a
five point scale. The specific ratings of our particular interest are the onset of corrosion
in the crevice (as evidenced by weeping of corrosion products), slight corrosion attack
around the crevice, and eventual structural deterioration. Within the USMC nomencla-
ture these are labeled as condition 2, 3, and 4 deterioration. Table 1 provides the data
for trailer crevice corrosion since the date of trailer manufacture.
11
5 1 4 1 17 6% 24% 6%
From this table it is possible to see that older trailers are more likely to have a higher
rating (higher degree of corrosion). As a high percentage of assets corrode enough to
become category 3 and 4, fewer assets are left as category 2. From this a timeline can
be created which states that by the end of year 1, 15% of the trailers will be assessed
as being in category 2.
Consider that the nominal shell thickness for these multiple classes of trailer may be in
the range of 50 to 100 mils and that in an unprotected condition 20+ mils of corrosion
can occur in 2 to 3 years, the occurrence of the loss of a structural joint (rating of 4) in
6% of the assets in 5 years is not surprising. The fact that it is not higher than 6% is
probably the positive impact of the coatings or other preservatives that were added over
the years. But again, if the failure point is to maintain structural integrity for 5+ years,
the corrosion rate data would have suggested that additional corrosion prevention is
needed to stop corrosion over this timeframe.
Knowing the percentage of the population that may achieve the failure condition (here
6%) the operating unit can then analyze the costs to improve the corrosion resistance of
all of its trailers to meet the five year requirement versus the cost to replace 6% of its
fleet every 5 years.
CONCLUSIONS
It is possible to predict coating life cycles through the creation of a timeline to failure. To
do so the following steps must be taken:
1. Set asset specific, objective failure criteria based on desired performance cha-
racteristics.
2. Develop a model of the corrosion process of interest which addresses the per-
formance characteristics of concern.
3. Obtain in-service data relating to the coating system and environment to which
the asset is going to be exposed. The data must reflect (measure) the perfor-
mance characteristics of interest (e.g., coating degradation and corrosion rate).
4. Use simple statistical modeling to predict the life span of the coating.
REFERENCES
12
3. Melchers, R.E. “Modeling of Marine Immersion Corrosion for Mild and Low-Alloy
Steels,” Corrosion, Vol. 59, No. 4, pp. 319-334, 2003
4. Paik, Thayamballi, Park, Hwang, “A Time-Dependent Corrosion Wastage Model for
Seawater Ballast Tank Structures of Ships”, Corrosion Science 46 (2004) pp. 471-486
5. Repp, J. & I. Handsy “Corrosion Evaluation of Army and Automotive Materials in
Hawaii – Year Three Update,”, SAE Conference, 2008-08B-298
6. Elzly Technology Corporation report to TACOM, “Engineering Support to the
PRCRP,” W56HZV-06-P-L647, September 2008
13