Thickness Monitoring
Content
ASSET INTEGRITY INTELLIGENCE
Fe atured Article
VOIDING 5 COMMON PITFALLS OF PRESSURE
A
VESSEL THICKNESS MONITORING
BY A.C. GYSBERS, THE EQUITY ENGINEERING GROUP
VOLUME 20, ISSUE 4
JULY | AUGUST 2014
AVOIDING 5 COMMON PITFALLS OF
PRESSURE VESSEL THICKNESS MONITORING
BY: A.C. GYSBERS
Corporate Principal Materials Engineer, The Equity Engineering Group
INTRODUCTION
One of the more common inspection monitoring programs
for pressure vessels is to perform thickness measurement at
Corrosion Monitoring Locations (CMLs) to allow monitoring of
minimum thicknesses and provide estimates for corrosion rates.
These minimum thicknesses and corrosion rates are critical in
supporting risk based inspection techniques or in setting half-life
prescriptive re-inspection intervals.
equipment strategy is well-defined (via a risk based assessment)
and appropriate thickness monitoring is identified for the new or
revamped equipment.
PITFALL #2: NOT COLLECTING BASELINE DATA
Often when reviewing the thickness database for pressure vessels, the installation thicknesses found at CMLs is nothing but
the nominal plate or component thickness at that location, as
it was considered that the new vessel had no need for baseline
thickness data and would be collected at the next appropriate
inspection interval. Unfortunately there can be a significant variance in actual plate or component thicknesses when installed,
and the ability to compare the first set of actual thickness readings against original readings to get a more accurate estimate of
metal losses and estimates for corrosion rates are lost for this first
processing period.
I have previously documented the problems that can arise when
performing thickness monitoring for piping in a series of articles
published in Inspectioneering Journal from 2012-2013 1,2,3,4,5. It
has been the author’s experience when reviewing plant inspection data for pressure vessels in preparation for risk based inspection assessments that the quality of thickness data and corrosion
rates for pressure vessels is often as much of a concern as piping
thickness data. This article will outline the common pitfalls assoHence, in conjunction with the initial assessment defining that
ciated with the pressure vessel thickness monitoring process and
thickness monitoring is required and defines where CMLs should
provide recommendations on correcting these.
be located and the type of corrosion expected, an initial set of
It is important to keep in mind that this article focuses only on actual thickness readings should be collected and entered into the
thinning, or corrosion, which in many plants, is the most preva- thickness database as the starting point for the vessel’s thickness
lent form of degradation. However, numerous other mechanisms monitoring program.
can lead to pressure vessel failures, such as fatigue, de-alloyPITFALL #3: THICKNESS MONITORING USING ONLY A
ing, environmentally assisted cracking (e.g. chloride SCC), etc.
SPOT CML APPROACH
Effective inspection strategies should be developed to detect the
All too often the thickness data entered into the pressure vessel
particular types of anticipated or potential damage as appropriate.
thickness database is collected as single point readings from individual CML locations that are equivalent to 2 to 3 inch diameter
PITFALL #1: LOCATING CMLS WITHOUT A
circles on the vessel or through the insulation of the vessel. Even
DEGRADATION ASSESSMENT
Many users establish CML locations on a new pressure vessel if the vessel was subject to only general corrosion, this type of
based on established plant or company standards. A vessel (tower, data gathering can lead to high data variances and a poor basis
drum, heat exchanger, etc.) sketch has a standard distribution of for establishing corrosion rates between the spot readings. In
CMLs located on the shell, heads, and nozzles. Unfortunately, this an article published in the November/December 2012 issue of
one-size-fits-all approach has little value in most circumstances, Inspectioneering Journal, I discussed how an actual corroding 5
as the actual corrosion anticipated in the vessel will be signifi- mpy corrosion rate using simple spot UT readings taken at differcantly different depending on service and the degradation mech- ent times could provide a calculated corrosion rate of anywhere
from -36 to 46 mpy based on typical variances associated with coranisms present.
rosion and the UT thickness measurement technique.
It would be more effective to have a materials or inspection specialist perform a degradation assessment in order to better estab- Hence if general corrosion is expected, the CML location may
lish the expected service for the vessel; if corrosion is a probable require multiple thickness monitoring locations and at each
degradation, he or she should be able to tell you what form of location, multiple individual thickness readings taken to estabcorrosion (general, local, pitting) is expected and what the prob- lish an average of three readings to be entered into the thickness
able location(s) for damage in the pressure vessel are. If the risk database. If individual thickness readings exceed the average
assessment warrants, CMLs can be appropriately concentrated in by more than ±0.030 inches or 10%, consideration has to be given
locations where the corrosion is anticipated and exercise the type to follow-up for either measurement error or some form of
of thickness monitoring at the CML that has been recommended localized corrosion.
for the type of degradation expected.
Often, the corrosion pattern may be localized to a particular
In highly effective programs, this is part of the project deliver- predicted location within a vessel. One common example is the
ables for new equipment or revamped equipment so that when inlet impingement area of a vessel or heat exchanger, which can
the equipment is installed and turned over to site personnel, the be exposed to a stream carrying sour water or acidic condensate,
JULY | AUGUST 2014
Inspectioneering Journal
29
Figure 1. Example of Close Grid UT TML Coverage (2 in. squares)
making it susceptible to local erosion/corrosion. Here the CMLs
should consist of multiple reasonably spaced TMLs arranged in
the expected impingement zone so that initial monitoring can
be undertaken externally for this potential. See Pitfall 4 for a discussion on the response to this type of monitoring for localized
corrosion issues after performing an internal inspection. Another
common example may be local corrosion at liquid levels in towers, such as at tray or downcomer levels where acid salts may collect. Here again, CMLs should be set up to with multiple TMLs to
cover the bandwidth expected at the tray levels located via mapping measurements on the external of the shell.
In many vessels or heat exchangers deposits and corroding species may collect along the bottom shell (horizontal) or heads
(vertical) and the most likely form of corrosion is underdeposit
pitting. External pitting corrosion thickness monitoring requires
a different consideration than the spot TML typically requires.
Pitting corrosion should be monitored on-stream using three
to five locations along the expected pitting zone via a larger
area CML (minimum 6 by 6 inch). Here, the area should be UT
scanned to detect and measure the pitting. The deepest pit should
be recorded (minimum thickness at the pit bottom) and stored in
your database for future reference and corrosion rate and remaining life assessments.
from the internal inspections. This can be used in most databases
to track and flag remaining life limitations for a particular vessel
and to support thickness based risk assessments. Depending on
the size and form of localized corrosion, the retirement thickness
basis at this TML can also be adjusted (½tret), as allowed by API
510, Pressure Vessel Inspection Code, to reflect pitting or small size
localized corrosion. As most databases use a measured thickness
as an entry point (and not corrosion depth), one should make sure
that the standard for collecting internal thickness data includes a
process for measuring the depth of corrosion (pit gauges, multiple readings) and the actual local thickness around the corroded
area to define by subtraction the estimate of minimum thickness
for data entry for the location, particularly if an on-stream monitoring CML is not created.
The next step in improving monitoring effectiveness would be
to actually create these CMLs externally on the vessel with the
capability to perform on-stream thickness monitoring at these
new locations. This is particularly necessary if the corrosion is
limiting the run length of the equipment relative to unit expectations, or to support internal inspection interval extensions. Of
course access/insulation considerations come into play when
considering on-stream monitoring and the risk mitigation
options and costs for alternative inspections (repeat internal
inspections verses on-stream) will determine if it is worthwhile
to use on-stream monitoring. If on-stream CML monitoring is
considered viable, it is important that it also be planned during
the internal inspection and that initial location and external
UT (spots, grids, scans) be completed while the equipment is
open to confirm that the on-stream CML locations represent the
internal findings.
It is important that this localized on-stream monitoring be performed by qualified NDE personnel utilizing proven procedures.
Diligent NDE and owner/user companies now perform training and demonstration testing to verify the capabilities of an
examiner. Demonstrations can include calibration, temperature
compensation, the ability to read drawings, preparing surfaces,
understanding the owner/user taxonomy, and reporting requirements. In addition, very effective programs include blind demonstration testing of samples with some thickness variation to
PITFALL #4: CML COVERAGE NOT ADJUSTED BASED
pre-qualify examiners. This should include a localized corrosion
ON INTERNAL INSPECTION FINDINGS
One of the most aggravating and common problems in preparing detection and measurement blind test, typically done on a comfor a risk based assessment of a piece of equipment is compar- ponent with well-known, very localized thin areas that are to be
ing the database of CML thicknesses and corrosion rates with the scanned and reported.
findings from internal vessel inspections and finding that they
PITFALL #5: LOCALIZED CORROSION THICKNESS
do not match. Very often the internal inspection may report local
MONITORING NOT CONSISTENTLY PERFORMED
corrosion or pitting problems with measured depths of losses, but
If the corrosion found during internal or external inspections is
for some reason, the CML database for the vessel has not been
localized or pitting-based type corrosion, single point CMLs are
updated to reflect these findings. The original CML pattern data
not very effective in ensuring accurate detection and measureis still in place and the data may not reflect the internal inspection
ment of the minimum thicknesses within the corroding zone. In
findings, making it difficult to establish a minimum thickness
the author’s opinion, the most effective technique for monitorand corrosion rate basis for remaining life and risk assessments.
ing localized corrosion is automated UT mapping with x-y raster
There are two steps to improve monitoring effectiveness to bet- scanning, which provides consistent minimum thickness identiter ensure that thickness monitoring data is consistent with the fication with the added advantage of mapping detail that could
internal inspection findings.
also support fitness-for-service analysis.
At a minimum, it is recommended that CML data entry(s) be cre- Often though, this technique may not be available and may
ated in the database to define the minimum thickness findings be costly. Another alternative and effective technique is using
30
Inspectioneering Journal
JULY | AUGUST 2014
Figure 2a. Spreadsheet Map of Raw UT Grid Data
Figure 2b. UT Grid Data (per 5a averages)
manual spot UT within the area of localized corrosion using a
close (2 inch) grid. The problem here is that there often are inconsistencies in collecting the data between surveys and a difficulty
in understanding what the real differences are between surveys
to estimate actual corrosion rates. Reference 3 provides guidelines
on how to take and analyze UT grid type thickness monitoring. To
improve repeatability and future comparisons, each grid square
can be treated as an individual TML and again, three individual
spot readings can be taken with each square (see Figure 1).
CONCLUSION
Though vessels have the advantage of having internal visual
inspections, thickness monitoring is still an essential part of confirming the remaining life of equipment and is critical for methodologies such as risk based inspection. Despite the apparent
simplicity of collecting thickness data, there are still many potential issues with gathering the appropriate data and properly analyzing the data. This article touched on some of these common
pitfalls in collecting and document thickness monitoring data for
pressure vessels to help the user maximize the effectiveness of
The same simple tests of the three readings discussed previ- their thickness data monitoring program. n
ously (individual reading > ±0.030 in. or > ±10% from the average
of the three reading) can be used for each grid TML to focus in
on grid locations that may even have highly localized corrosion. References
1. Gysbers, A.C., 2012, “A Discussion on the Piping Thickness Management ProSimple spreadsheet data plotting provides a quick visual map of
cess”, Inspectioneering Journal, 18(5), pp. 10-13.
the survey data for testing the local TML variance (Figure 2a), 2. Gysbers, A.C., 2012, “A Discussion on the Piping Thickness Management Process: Part 2 – Determining Corrosion Monitoring Locations”, Inspectioneering
and plotting the overall data using local averages (Figure 2b)
Journal, 18(6), pp. 6-11.
not only documents the losses, but provides a repeatable map 3.Gysbers, A.C., 2013, “A Discussion on the Piping Thickness Management Proto reduce variance for future comparison when the survey is cess: Part 3 - Data Collection with Ultrasonics and Radiography”, Inspectioneering Journal, 19(1), pp. 11-15.
performed again.
4. Gysbers, A.C., 2013, “A Discussion on the Piping Thickness Management ProThese surveys may help determine where the current thinnest
location is positioned and pinpoint a single TML for future
monitoring. However, I would not recommend this approach,
as localized corrosion can be highly inconsistent and change
locally during operations. The repeated area can be reduced
from the initial survey, but should still consist of a grid at and
around the thinned area to confirm that the pattern has not significantly changed.
cess: Part 4 - Collecting Quality Thickness Data”, Inspectioneering Journal, 19(3),
pp. 12-14.
5. Gysbers, A.C., 2013, “A Discussion on the Piping Thickness Management Process: Part 5 - Circuit Thickness Data”, Inspectioneering Journal, 19(5), pp. 8-15.
READ THIS ARTICLE ON
INSPECTIONEERING.COM
JULY | AUGUST 2014
Inspectioneering Journal
31
Fe atured Article
VOIDING 5 COMMON PITFALLS OF PRESSURE
A
VESSEL THICKNESS MONITORING
BY A.C. GYSBERS, THE EQUITY ENGINEERING GROUP
VOLUME 20, ISSUE 4
JULY | AUGUST 2014
AVOIDING 5 COMMON PITFALLS OF
PRESSURE VESSEL THICKNESS MONITORING
BY: A.C. GYSBERS
Corporate Principal Materials Engineer, The Equity Engineering Group
INTRODUCTION
One of the more common inspection monitoring programs
for pressure vessels is to perform thickness measurement at
Corrosion Monitoring Locations (CMLs) to allow monitoring of
minimum thicknesses and provide estimates for corrosion rates.
These minimum thicknesses and corrosion rates are critical in
supporting risk based inspection techniques or in setting half-life
prescriptive re-inspection intervals.
equipment strategy is well-defined (via a risk based assessment)
and appropriate thickness monitoring is identified for the new or
revamped equipment.
PITFALL #2: NOT COLLECTING BASELINE DATA
Often when reviewing the thickness database for pressure vessels, the installation thicknesses found at CMLs is nothing but
the nominal plate or component thickness at that location, as
it was considered that the new vessel had no need for baseline
thickness data and would be collected at the next appropriate
inspection interval. Unfortunately there can be a significant variance in actual plate or component thicknesses when installed,
and the ability to compare the first set of actual thickness readings against original readings to get a more accurate estimate of
metal losses and estimates for corrosion rates are lost for this first
processing period.
I have previously documented the problems that can arise when
performing thickness monitoring for piping in a series of articles
published in Inspectioneering Journal from 2012-2013 1,2,3,4,5. It
has been the author’s experience when reviewing plant inspection data for pressure vessels in preparation for risk based inspection assessments that the quality of thickness data and corrosion
rates for pressure vessels is often as much of a concern as piping
thickness data. This article will outline the common pitfalls assoHence, in conjunction with the initial assessment defining that
ciated with the pressure vessel thickness monitoring process and
thickness monitoring is required and defines where CMLs should
provide recommendations on correcting these.
be located and the type of corrosion expected, an initial set of
It is important to keep in mind that this article focuses only on actual thickness readings should be collected and entered into the
thinning, or corrosion, which in many plants, is the most preva- thickness database as the starting point for the vessel’s thickness
lent form of degradation. However, numerous other mechanisms monitoring program.
can lead to pressure vessel failures, such as fatigue, de-alloyPITFALL #3: THICKNESS MONITORING USING ONLY A
ing, environmentally assisted cracking (e.g. chloride SCC), etc.
SPOT CML APPROACH
Effective inspection strategies should be developed to detect the
All too often the thickness data entered into the pressure vessel
particular types of anticipated or potential damage as appropriate.
thickness database is collected as single point readings from individual CML locations that are equivalent to 2 to 3 inch diameter
PITFALL #1: LOCATING CMLS WITHOUT A
circles on the vessel or through the insulation of the vessel. Even
DEGRADATION ASSESSMENT
Many users establish CML locations on a new pressure vessel if the vessel was subject to only general corrosion, this type of
based on established plant or company standards. A vessel (tower, data gathering can lead to high data variances and a poor basis
drum, heat exchanger, etc.) sketch has a standard distribution of for establishing corrosion rates between the spot readings. In
CMLs located on the shell, heads, and nozzles. Unfortunately, this an article published in the November/December 2012 issue of
one-size-fits-all approach has little value in most circumstances, Inspectioneering Journal, I discussed how an actual corroding 5
as the actual corrosion anticipated in the vessel will be signifi- mpy corrosion rate using simple spot UT readings taken at differcantly different depending on service and the degradation mech- ent times could provide a calculated corrosion rate of anywhere
from -36 to 46 mpy based on typical variances associated with coranisms present.
rosion and the UT thickness measurement technique.
It would be more effective to have a materials or inspection specialist perform a degradation assessment in order to better estab- Hence if general corrosion is expected, the CML location may
lish the expected service for the vessel; if corrosion is a probable require multiple thickness monitoring locations and at each
degradation, he or she should be able to tell you what form of location, multiple individual thickness readings taken to estabcorrosion (general, local, pitting) is expected and what the prob- lish an average of three readings to be entered into the thickness
able location(s) for damage in the pressure vessel are. If the risk database. If individual thickness readings exceed the average
assessment warrants, CMLs can be appropriately concentrated in by more than ±0.030 inches or 10%, consideration has to be given
locations where the corrosion is anticipated and exercise the type to follow-up for either measurement error or some form of
of thickness monitoring at the CML that has been recommended localized corrosion.
for the type of degradation expected.
Often, the corrosion pattern may be localized to a particular
In highly effective programs, this is part of the project deliver- predicted location within a vessel. One common example is the
ables for new equipment or revamped equipment so that when inlet impingement area of a vessel or heat exchanger, which can
the equipment is installed and turned over to site personnel, the be exposed to a stream carrying sour water or acidic condensate,
JULY | AUGUST 2014
Inspectioneering Journal
29
Figure 1. Example of Close Grid UT TML Coverage (2 in. squares)
making it susceptible to local erosion/corrosion. Here the CMLs
should consist of multiple reasonably spaced TMLs arranged in
the expected impingement zone so that initial monitoring can
be undertaken externally for this potential. See Pitfall 4 for a discussion on the response to this type of monitoring for localized
corrosion issues after performing an internal inspection. Another
common example may be local corrosion at liquid levels in towers, such as at tray or downcomer levels where acid salts may collect. Here again, CMLs should be set up to with multiple TMLs to
cover the bandwidth expected at the tray levels located via mapping measurements on the external of the shell.
In many vessels or heat exchangers deposits and corroding species may collect along the bottom shell (horizontal) or heads
(vertical) and the most likely form of corrosion is underdeposit
pitting. External pitting corrosion thickness monitoring requires
a different consideration than the spot TML typically requires.
Pitting corrosion should be monitored on-stream using three
to five locations along the expected pitting zone via a larger
area CML (minimum 6 by 6 inch). Here, the area should be UT
scanned to detect and measure the pitting. The deepest pit should
be recorded (minimum thickness at the pit bottom) and stored in
your database for future reference and corrosion rate and remaining life assessments.
from the internal inspections. This can be used in most databases
to track and flag remaining life limitations for a particular vessel
and to support thickness based risk assessments. Depending on
the size and form of localized corrosion, the retirement thickness
basis at this TML can also be adjusted (½tret), as allowed by API
510, Pressure Vessel Inspection Code, to reflect pitting or small size
localized corrosion. As most databases use a measured thickness
as an entry point (and not corrosion depth), one should make sure
that the standard for collecting internal thickness data includes a
process for measuring the depth of corrosion (pit gauges, multiple readings) and the actual local thickness around the corroded
area to define by subtraction the estimate of minimum thickness
for data entry for the location, particularly if an on-stream monitoring CML is not created.
The next step in improving monitoring effectiveness would be
to actually create these CMLs externally on the vessel with the
capability to perform on-stream thickness monitoring at these
new locations. This is particularly necessary if the corrosion is
limiting the run length of the equipment relative to unit expectations, or to support internal inspection interval extensions. Of
course access/insulation considerations come into play when
considering on-stream monitoring and the risk mitigation
options and costs for alternative inspections (repeat internal
inspections verses on-stream) will determine if it is worthwhile
to use on-stream monitoring. If on-stream CML monitoring is
considered viable, it is important that it also be planned during
the internal inspection and that initial location and external
UT (spots, grids, scans) be completed while the equipment is
open to confirm that the on-stream CML locations represent the
internal findings.
It is important that this localized on-stream monitoring be performed by qualified NDE personnel utilizing proven procedures.
Diligent NDE and owner/user companies now perform training and demonstration testing to verify the capabilities of an
examiner. Demonstrations can include calibration, temperature
compensation, the ability to read drawings, preparing surfaces,
understanding the owner/user taxonomy, and reporting requirements. In addition, very effective programs include blind demonstration testing of samples with some thickness variation to
PITFALL #4: CML COVERAGE NOT ADJUSTED BASED
pre-qualify examiners. This should include a localized corrosion
ON INTERNAL INSPECTION FINDINGS
One of the most aggravating and common problems in preparing detection and measurement blind test, typically done on a comfor a risk based assessment of a piece of equipment is compar- ponent with well-known, very localized thin areas that are to be
ing the database of CML thicknesses and corrosion rates with the scanned and reported.
findings from internal vessel inspections and finding that they
PITFALL #5: LOCALIZED CORROSION THICKNESS
do not match. Very often the internal inspection may report local
MONITORING NOT CONSISTENTLY PERFORMED
corrosion or pitting problems with measured depths of losses, but
If the corrosion found during internal or external inspections is
for some reason, the CML database for the vessel has not been
localized or pitting-based type corrosion, single point CMLs are
updated to reflect these findings. The original CML pattern data
not very effective in ensuring accurate detection and measureis still in place and the data may not reflect the internal inspection
ment of the minimum thicknesses within the corroding zone. In
findings, making it difficult to establish a minimum thickness
the author’s opinion, the most effective technique for monitorand corrosion rate basis for remaining life and risk assessments.
ing localized corrosion is automated UT mapping with x-y raster
There are two steps to improve monitoring effectiveness to bet- scanning, which provides consistent minimum thickness identiter ensure that thickness monitoring data is consistent with the fication with the added advantage of mapping detail that could
internal inspection findings.
also support fitness-for-service analysis.
At a minimum, it is recommended that CML data entry(s) be cre- Often though, this technique may not be available and may
ated in the database to define the minimum thickness findings be costly. Another alternative and effective technique is using
30
Inspectioneering Journal
JULY | AUGUST 2014
Figure 2a. Spreadsheet Map of Raw UT Grid Data
Figure 2b. UT Grid Data (per 5a averages)
manual spot UT within the area of localized corrosion using a
close (2 inch) grid. The problem here is that there often are inconsistencies in collecting the data between surveys and a difficulty
in understanding what the real differences are between surveys
to estimate actual corrosion rates. Reference 3 provides guidelines
on how to take and analyze UT grid type thickness monitoring. To
improve repeatability and future comparisons, each grid square
can be treated as an individual TML and again, three individual
spot readings can be taken with each square (see Figure 1).
CONCLUSION
Though vessels have the advantage of having internal visual
inspections, thickness monitoring is still an essential part of confirming the remaining life of equipment and is critical for methodologies such as risk based inspection. Despite the apparent
simplicity of collecting thickness data, there are still many potential issues with gathering the appropriate data and properly analyzing the data. This article touched on some of these common
pitfalls in collecting and document thickness monitoring data for
pressure vessels to help the user maximize the effectiveness of
The same simple tests of the three readings discussed previ- their thickness data monitoring program. n
ously (individual reading > ±0.030 in. or > ±10% from the average
of the three reading) can be used for each grid TML to focus in
on grid locations that may even have highly localized corrosion. References
1. Gysbers, A.C., 2012, “A Discussion on the Piping Thickness Management ProSimple spreadsheet data plotting provides a quick visual map of
cess”, Inspectioneering Journal, 18(5), pp. 10-13.
the survey data for testing the local TML variance (Figure 2a), 2. Gysbers, A.C., 2012, “A Discussion on the Piping Thickness Management Process: Part 2 – Determining Corrosion Monitoring Locations”, Inspectioneering
and plotting the overall data using local averages (Figure 2b)
Journal, 18(6), pp. 6-11.
not only documents the losses, but provides a repeatable map 3.Gysbers, A.C., 2013, “A Discussion on the Piping Thickness Management Proto reduce variance for future comparison when the survey is cess: Part 3 - Data Collection with Ultrasonics and Radiography”, Inspectioneering Journal, 19(1), pp. 11-15.
performed again.
4. Gysbers, A.C., 2013, “A Discussion on the Piping Thickness Management ProThese surveys may help determine where the current thinnest
location is positioned and pinpoint a single TML for future
monitoring. However, I would not recommend this approach,
as localized corrosion can be highly inconsistent and change
locally during operations. The repeated area can be reduced
from the initial survey, but should still consist of a grid at and
around the thinned area to confirm that the pattern has not significantly changed.
cess: Part 4 - Collecting Quality Thickness Data”, Inspectioneering Journal, 19(3),
pp. 12-14.
5. Gysbers, A.C., 2013, “A Discussion on the Piping Thickness Management Process: Part 5 - Circuit Thickness Data”, Inspectioneering Journal, 19(5), pp. 8-15.
READ THIS ARTICLE ON
INSPECTIONEERING.COM
JULY | AUGUST 2014
Inspectioneering Journal
31
