Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
CONTENTS
O.1. GENERAL ......................................................................................................................................................8
O 1.1 Scope .....................................................................................................................................................8
O 1.2 References .............................................................................................................................................8
O 1.3 Definitions .............................................................................................................................................9
O.2. INTRODUCTION...........................................................................................................................................9
O 2.1 RBI Methodology Introduction .............................................................................................................9
O 2.2 Background..........................................................................................................................................13
O 2.3 Generic Failure Frequencies (GFF) – AST Floor and Shell ................................................................14
O.2.3.1 GFF – Floor Failures.......................................................................................................................15
O.2.3.2 GFF – Shell Failure.........................................................................................................................16
O 2.4 Advantages with RBI for AST’s..........................................................................................................16
O 2.5 Limitations, Assumptions, Guidelines and Recommendations............................................................17
O.2.5.1 Assumptions....................................................................................................................................17
O.2.5.2 Limitations ......................................................................................................................................18
O.2.5.3 Risk Profile Plot..............................................................................................................................18
O.2.5.4 General Guidelines..........................................................................................................................19
O.2.5.5 Conservatism Built into the RBI Methodology ..............................................................................23
O.3. METHODOLOGY OVERVIEW .................................................................................................................25
O.3.1.1 Risk .................................................................................................................................................26
O.3.1.2 Likelihood of Failure ......................................................................................................................26
O.3.1.3 Consequence ...................................................................................................................................26
O.3.1.4 Relative Risk...................................................................................................................................27
O.3.1.5 Total Risk – Tank Specific .............................................................................................................28
O.3.1.6 Risk Limits and Inspection Planning ..............................................................................................29
O.4. TANK BOTTOM – LIKELIHOOD OF FAILURE......................................................................................32
O 4.1 Scope – Leak and Rupture...................................................................................................................32
O 4.2 Methodology Overview – Leak and Rupture ......................................................................................32
O 4.3 Input Data – Leak and Rupture............................................................................................................34
O 4.4 Corrosion Rate Calculation – Leak Failure .........................................................................................37
O.4.4.1 Soil Side Corrosion Rate.................................................................................................................39
O.4.4.2 Product Side Corrosion Rate...........................................................................................................41
O.4.4.3 Determine combined Tank Floor Corrosion Rate ...........................................................................44
O 4.5 Measured Wall Thickness and Corrosion Rate....................................................................................45
O.4.5.1 Re-Set RBI Analysis – New Wall Thickness..................................................................................45
FIGURES:
Figure 1: Overview of Main Tasks..............................................................................................................................10
Figure 2: Analysis Calculation Overview....................................................................................................................11
Figure 3: Overview of main methodology elements and related details needed for these. .........................................12
Figure 4: Five-By-Five Risk Matrix ............................................................................................................................27
Figure 5: Risk Matrix Showing Levels of Risk ...........................................................................................................27
Figure 6: Risk Level Calculation.................................................................................................................................28
Figure 7: Risk Rank Matrix (LoF and CoF Categories) and Recommend DFTargets ....................................................30
Figure 8: Risk Target and Inspection Planning ...........................................................................................................31
Figure 9: Likelihood of Failure Calculation for Tank Floor........................................................................................33
Figure 10: Calculation of Corrosion Rate (r) to derive Modification Factor (DFar/t)........................................................38
Figure 11: Revised MFLiner ..........................................................................................................................................51
Figure 12: DF & LoF Calculation for Tank Shell .......................................................................................................55
Figure 13 : Tank Shell CR observed with Shell Height ..............................................................................................67
Figure 14: CR Index Ratio to Course 1 - as a Function of Total Number of Courses.................................................68
Figure 15: Estimated CR in H2S – Air 8, See Reference /7/ .......................................................................................70
Figure 16: Changes in Corrosion Rates with Slow and Rapid Liquid Line Motion....................................................72
Figure 17: Flow Chart to determine estimated corrosion rates for Welded Shells ......................................................74
Figure 18: Climate Map for the United States.............................................................................................................77
Figure 19: Illustration of the calculated Risk Ratio.....................................................................................................83
Figure 20: Overview of Spill Consequence Model .....................................................................................................90
Figure 21: Flowchart for Consequence Model ............................................................................................................93
Figure 22: Environmental Consequence Calculation – Bottom Leak .........................................................................96
Figure 23: Environmental Consequence Calculation – Bottom Rupture.....................................................................97
Figure 24: Direct Financial Consequence Calculation ................................................................................................99
Figure 25: Floor Leakage Illustration..........................................................................................................................99
Figure 26: Flow Out of Tank Hole ............................................................................................................................104
Figure 27: Vertical Fluid Velocity.............................................................................................................................105
Figure 28: Flow Out of RPB .....................................................................................................................................106
Figure 29: Release from Tank Shell ..........................................................................................................................107
Figure 30: CoF Calculation .......................................................................................................................................117
Figure 31: DF Calculation .........................................................................................................................................118
Figure 32: LoF Calculation .......................................................................................................................................119
TABLES:
Table 1: Liquid release scenarios analyzed in the AST RBI methodology .................................................................13
Table 2: Base Leak Frequencies for Tank Floor and Shell .........................................................................................14
Table 3: Summary of Survey Results ..........................................................................................................................15
Table 4: Default – Recommended Inspection Targets.................................................................................................25
Table 5: Input Data Required for Bottom Corrosion Rate Calculation .......................................................................34
Table 6: Input Data Required for DF and Likelihood Calculation ..............................................................................35
Table 7: Input Data Required for Consequence Calculation .......................................................................................36
Table 8: Input Data for Risk evaluation and Inspection Planning..................................................................................37
Table 9: Summary of Conditions for Soil Side Base Corrosion Rate of 5mpy ...........................................................39
Table 10: Soil Resistivity ............................................................................................................................................39
Table 11: Tank Pad Adjustment ..................................................................................................................................40
Table 12: Tank Drainage Adjustment .........................................................................................................................40
Table 13: Adjustment for Cathodic Protection............................................................................................................41
Table 14: Adjustment for Bottom Type.......................................................................................................................41
Table 15: Adjustment for Fluid Temperature..............................................................................................................41
Table 16: Product Side Base Corrosion Rates.............................................................................................................42
Table 17: Summary of Conditions for ‘Base’ Product Side Corrosion Rate...............................................................42
Table 18: Steam Coil Heater Adjustment ....................................................................................................................42
Table 19: Water Draw Adjustment..............................................................................................................................42
Table 20: Internal Coating Needed?............................................................................................................................43
Table 21: Coating Quality ...........................................................................................................................................43
Table 22: Correction Factor (CWT and CCR) for measured Floor thickness readings and CR......................................45
Table 23: Guidelines for Assigning Inspection Rating – Tank Bottom ......................................................................47
Table 24: Tank Bottom DFar/t Damage Factors .........................................................................................................49
Table 25: Damage Factor for Tank Design and Maintenance .....................................................................................53
Table 26: Damage Factor for Tank Settlement............................................................................................................54
Table 27: Basic Data Required for Shell RBI Analysis...............................................................................................56
Table 28: Brittle Fracture Susceptibility – Screening Questions.................................................................................61
Table 29: Cracking of Low Alloy Steel and Carbon Steel Tank Shells – Answer Q1 OR Q2....................................63
EQUATIONS:
Equation 1: Risk Calculation.......................................................................................................................................26
Equation 2: LoF Calculation........................................................................................................................................26
Equation 3: CoF Calculation .......................................................................................................................................26
Equation 4: CUF Caclulation ......................................................................................................................................27
Equation 5: Coating Factor Calculation ......................................................................................................................43
Equation 6; Measured CR adjsutement .......................................................................................................................46
Equation 7: ar/t ............................................................................................................................................................48
Equation 8: Bottom Leak Failure Probability..............................................................................................................50
Equation 9: Liner Modification Factor ........................................................................................................................51
Equation 10: Rapid Bottom Failure Probability ..........................................................................................................53
O.1. GENERAL
O 1.1 Scope
The scope of this document is to provide a methodology for planning inspection of aboveground storage
tanks (AST) based on environmental risk, and complies with the principles of API 580 in terms of Risk
Based Inspection.
The methodology applies to atmospheric AST’s, designed to API 12D or API 650, and used for storage of
hydrocarbon fluids. The fluid types mainly relate to the consequence calculations1.
A qualitative screening method is provided for brittle fracture assessment and evaluation of potential
cracking damage mechanism on the tank shell. The screening output is a qualitative level of susceptibility
to brittle fracture and cracking, and the mechanisms by which cracking is likely to be driven.
A quantitative method is provided for potential corrosion damage on both the tank shell and the tank
floor. Widespread corrosion and localized corrosion, including pitting and erosion-corrosion, are within
the scope of this methodology, and address internal and external corrosion on both shell and floor
surfaces.
O 1.2 References
The following references material has been used in developing this RBI methodology:
/1/ Aboveground Storage Tank, Risk Assessment Manual (RAM), API Aboveground Storage Tank
Committee, January 2002 (draft Appendix A and B issued for ballot January 2002)
/2/ Welded Steel Tanks for Oil Storage, API 650, 10th Edition, November 1998, including
Addendum 1 January 2000 and Addendum 2, November 2001.
/3/ Cathodic Protection of Aboveground Petroleum Storage Tanks, API RP 651, 2nd Edition,
November 1997
/4/ Lining of Aboveground Petroleum Storage Tank Bottoms, API RP 652, 2nd Edition, December
1997
/5/ Tank Inspection, Repair, Alteration and Reconstruction, API 653, 3rd Edition, December 2001
/6/ ASM, Metals Handbook, Ninth Edition, Vol. 13 Corrosion (1987).
/7/ Dr. Richard A. White, Materials Selection for Petroleum Refineries and Gathering Facilities,
NACE International, 1998, Houston, Texas
/8/ Inspection of Atmospheric and Low-Pressure Storage Tanks, API RP 575, 1st Edition, November
1995
/9/ Risk Based Inspection, API RP 580, 1st Edition, May 2002
/10/ Specification for Field Welded Tanks for Storage of Production Liquids, API 12D, 10th Edition,
November 1994
1
The original intent with this document was to address AST’s that are used for storage of hydrocarbon products –
the same as the API RAM document. However, with minor adjustments to some of the elements of the RBI
methodology, other type of fluids can also be considered. This is noted where applicable.
O 1.3 Definitions
*** EXPAND DURING REVIEW ***
O.1.4.1 damage factor (DF): An expression for the condition that a component is expected to be in, i.e.
it determines how close the component is to a theoretical failure. The DF is a function of the components
age (a), rate of corrosion (r), nominal wall thickness (t), and amount of past inspection. The DF is
determined from a look-up table by entering the ar/t value and picked the DF value from the appropriate
inspection effectiveness column. As expected, the variable is age, and the ar/t value is the theoretical
fraction of wall loss at the any given Age.
O.1.4.2 risk profile: The change of risk over time, and is in context of this document understood as the
change in damage factor (DF) over time. The profile can also be expressed in dollars/yr by multiplying
the DF by the probability weighted consequence (which is a fixed dollar value). However, depending on
the consequence, this will just scale up the profile, and with
O.1.4.3 damage factor inspection target (DFTarget): The tolerable limit for the damage factor. Once
the risk profile reaches the DF target an Inspection is recommended. The effectiveness of the inspection
should reduce the damage factor as much as possible, not only below the target, e.g. the highest attainable
inspection effectiveness is always recommended, provided it is practicable and cost effective.
O.2. INTRODUCTION
The objective with this document is to define a methodology for modeling risk associated with
Aboveground Storage tanks (AST’s), and how to apply a Risk Based approach for planning inspection of
AST’s. An ingrained goal is to determine the practical implementation of the RBI approach and to
establish a basis for validating the usage of this methodology in lieu of time based inspection planning
method.
AST inspection intervals can range widely (from a few years up to 20 years), depending on local
jurisdiction and company inspection philosophy. Likewise can the associated cost in terms of monetary
investment and increased hazards for environment and personnel, be significant. This risk prediction
model has been developed in search of the optimum timing for inspection based on the predicted risk, and
at the same time considers the companies (or local jurisdictions) tolerance towards risk. The objective
with this model is also to visualize and portray the risk progression for each AST as a function of time.
The most appropriate time for the next inspection is determined in relation to a defined risk criterion –
representing the risk tolerance. This in principle constitutes the inspection-planning task.
The deliverables for an AST RBI study should enable the owner/operator to present the theory and
practical application of the methodology to the local regulators, and prompt a debate about the basis on
which future AST internal inspections can be determined upon, e.g. RBI versus Time Based Intervals.
This is a local issue that needs to be raised on a case-to-case basis, and does not enter into the details of
this methodology document.
O 2.1 RBI Methodology Introduction
The content of this document is envisaged being used as a reference and guideline, and to ensure a
consistent approach for implementing RBI for AST’s. Unless the outlined methodology is supported by a
fairly rigorous convention for its implementation, in theory the analysis output could be manipulated to
show any desired result, and possibly rendering the methodology deceitful.
As illustrated very simplistically on Figure 1, there are only a few tasks involved in doing the RBI study
for an AST. The majority of the work relates to the data gathering.
The Cost benefit assessment task is added in parenthesis because this can be done in several different
ways and does not constitute an integral part of the RBI methodology. No Cost Benefit Assessment
method has been included in this report.
The detailed steps involved in performing the calculation and analysis are somewhat more complex than
implied on Figure 1. The details of the analysis task are shown on Figure 2.
It is suggested that the RBI program be applied conservatively, in order to align existing procedures and
practices with the RBI methodology, and ultimately become more confident with the whole approach.
Figure 2 provides a coarse overview of the elements of the methodology and the details required when
performing the analysis.
Figure 3 outlines the main elements that enter into the RBI methodology, and its intermediate analysis
steps and results.
Start Analysis
CoF for the Floor A CR for the Shell Courses CoF for the Shell B
CoF Category CR for the Floor Damage susceptibility Safe Filling Height CoF Category
assessment for the Shell
(cracking and brittle
Inspection Target fracture)
DF for the Floor
DF for the Shell Courses Inspection Target
A Risk related to the Floor Total AST Risk (Sum) Risk related to the Shell B
Figure 3: Overview of main methodology elements and related details needed for these.
Star
t
O 2.2 Background
Since around 1995, API has lead a committee to address issues relating to aboveground storage tanks,
with the objective to establish a recommended practice for risk assessment of AST’s. This has resulted in
a risk assessment manual (RAM), including one main report and two Appendices (‘A’ for Frequency
Analysis, and ‘B’ for Consequence Analysis). A draft of the RAM document was prepared in January
2002 for a formal ballot by the API AST group members. This very version of the RAM document is the
referred document in context of this methodology document (see reference /1/).
The objective with the API RAM project is to enable the user to determine practically all elements of risk
associated with AST operations at any one given point in time. The RAM addresses tank shell, bottom,
auxiliary piping, filling operation, etc. with the purpose of determining potential areas of concern.
Mitigation in context with the RAM could be inspection, improved operating procedures, equipment
layout, emergency response strategies, etc.
The Risk Based Inspection methodology that is defined in this RBI document does to a large extent build
on the principle and details of the API RAM document. The main difference is that the RBI methodology
is focused on inspection planning, by assessing risk as a function of time. Hence, determining an
appropriate time for inspection based on risk. The RBI approach considers risk associated with tank floor
and shell failures as drivers for inspection and or consequence mitigation, ref. Table 1.
The risk associated with the relevant failure scenarios in Table 1 is expressed as:
Risk = Likelihood × Consequence
where,
Relative
Total Comparison Frequency
Failure Scenario (Hole Size) Comparison
(Floor & Shells) (per year)
(GFF Relative Ratio)
The distribution does not consider the relative number of tanks in each group, i.e. welded versus riveted
or maintained to API 653 versus not maintained to API 653.
2
From a statistical standpoint, all tank Shells are initially assumed to have been designed to a recognized code, and
currently maintained to API 653. If not, the likelihood of a rupture failure is assumed a factor of 40 higher.
3
Rapid failures are not considered relevant for riveted tanks.
A bottom leak frequency of 7.2 × 10-3 leaks per year was chosen as the base leak frequency by the AST
RBI methodology. Although the leak frequency survey indicates that tanks less than 5 years old have a
much lower leak frequency, it has been decided to use the whole survey population in setting the base
leak frequency. The age of the tank is elsewhere accounted for in the methodology since the percent wall
loss is a function of the tank age, corrosion rate, and original wall thickness. The percent wall loss is the
basis of the modifier on the base leak frequency. Thus a very young tank with minimal corrosion will
have a frequency modifier that does not drive the predicted risk to the point where an inspection is
recommended.
The aforementioned survey did not report the size of leaks, but a survey of the sponsors for the API AST
RAP project indicated that leak sizes of ≤½ in. in diameter would adequately describe the vast majority of
tank bottom leaks.
b. 1988 Ashland Oil, PA, brittle fracture caused loss of 96,000 bbl diesel
One could argue that this set of data may be incomplete. It is difficult to imagine, however, that a
catastrophic failure of a large storage tank could escape the attention of the national news media.
The number of tanks that provided the basis for the two failures was estimated from literature to be about
33,300 large storage tanks. This value was based on a 1989 study carried out for API by Entropy Ltd.
Large, in this case, is defined as having a capacity greater than 10,000 barrels. The number of tanks
represents the total in the Unites States for the refining, marketing, transportation, and production sectors.
Thus, the total number of tank years was found to be approximately 1,000,000.
Dividing the number of failures by the number of tank years yields a rapid shell failure frequency of
2×10-6 per tank year. API Standard 653 requires an evaluation of tanks for susceptibility to brittle fracture.
Hydrostatic testing or re-rating of the tank may be required for continued service. API 653 thus provides
considerable protection against brittle fracture. Assuming one-half of the tanks are maintained to API 653,
then the base leak frequencies for rapid shell failures would be as shown in Table 2.
It has been assumed that rupture failure does not apply for riveted storage tanks.
O 2.4 Advantages with RBI for AST’s
Under normal circumstances the AST operator is faced with two methods or options to plan for internal
AST inspection, e.g. API 653 or a state regulated time based inspection frequency. As per the 1999
addendum to API 653, the code allows the owner-operator to establish internal inspection intervals using
RBI procedures, which in principle introduces a third option.
The application of a RBI program can be argued to introduce a basis for better and more pertinent
planning for AST inspection. A couple of items that are motivated by the implementation of RBI can be
expected to be:
a. Systematic categorization; All tanks are prioritized into formal Risk categories, and modeled as
individual components in the RBI analysis system. This creates a uniform and systematic base for
planning the inspection. The system is based on:
1) Consistent approach for calculating a representative value for tank floor degradation at any given
point in time, i.e. a quantitative assessment of the likelihood of failure.
2) Consistent approach for assessing the consequence of failure (CoF) for each specific tank, and
quantifying the CoF as well as risk in dollars per years.
b. Detailed assessment of all tanks; review of all tank data records, and assessment of inspection results
in context with the risk of failure over long periods of time, is part of the implementation process for
the RBI program.
c. Focus inspection effort to reduce risk; since the operating, maintenance and inspection history are
important inputs for analyzing a tank, RBI motivates a better focus on establishing good inspection
records that address the need for improving the AST model for each tank. This again should improve
the specific scope for each inspection that is planned.
d. Improved record keeping; The program should encourage the owner-operator to accumulate the
operational experience, which in time will prove valuable for enhancing the RBI program, as well as
future analysis work.
e. Enable long term planning and avoiding bottlenecks in terms of reducing inspection today, which
might result in an increased amount of inspection in 10 to 15 years from now.
b. The damage factor assumes that the thinning mechanism has resulted in a constant rate of
thinning/pitting over the period defined for the analysis. The likelihood of failure is estimated by
examining the possibility that the corrosion rate is greater than expected. The number and type of
inspections that have been performed determine the likelihood of discovering these higher rates. The
more thorough the inspection and the greater the number of inspections, the less likely is the chance
that the corrosion rate is greater than anticipated.
c. Repair of tank floor means restoring the minimum wall thickness to at least 30% of the original wall
thickness.
Assumptions related to specific adjustment factors or details of the methodology are mentioned in that
context.
O.2.5.2 Limitations
a. The module is only applicable for aboveground atmospheric storage tanks constructed in carbon steel
(floor and shell)
b. Consequence does not consider toxicity and fatality issues – only environmental and financial
impacts from failures.
c. Double floor tanks are considered as single bottom with a release prevention barrier. This is probably
too conservative when calculating the consequence of failure.
d. Not considering bottom design (i.e. cone up or cone down or flat bottom). Data is not currently
available to determine the impact of bottom design on failure frequencies (GFF) or corrosion rates.
e. Annular Ring is not considered separately. The ring is seen as an integral part of the floor, and
normally with a larger wall thickness. Based on experience, there is no reason to suppose that the
annular ring will fail, in terms of leakage, more frequently than other parts of the Tank floor.
However, the center section of the floor seems to be somewhat less prone to severe corrosion than
other part of the floor, in particular the floor perimeter.
f. Floor bulging or buckling – due to thermal cycling or varying filling height – is not specifically
modeled. The immediate impact from this scenario is that the Cathodic Protection (CP) system will
become ineffective when the floor is no longer in intimate contact with the pad underneath. Also the
risk of floor weld cracking increases significant if floor plates are overlay welded as oppose to butt-
welded, and the floor bulges due to thermal expansion. For operating temperature <250°F, bulging is
assumed not to be an issue.
b. Any major changes such as installing new floor, new sand pad, adding water draw, adding steam coil,
etc.
d. Change of service.
f. When an inspection is made. The objective is to give credit for individual inspections from the time
they are made, and reflect this on the risk profile. It is optional to reset the analysis (or WT), based on
good measurement results.
b. Cathodic protection; assume all cp systems comply with API RP 651. No credit is given for the cp
system until a highly effective inspection verifies that the system is indeed effective in arresting
corrosion.
c. Corrosion rate – calculated vs. measured; measured rates should be used whenever available,
otherwise the calculated rate should be used. Note, when the measured rates are used, the settings
contributing to a calculated rate (such as soil resistivity, CP system, etc.) Are not taken into account.
As part of the analysis and where relevant, the available measured cr should be compared to the
calculated CR, and in case there is a significant difference (≥25%), the CR should be subject to
further evaluation.
d. Corrosion Rate – measured; In case of hole through; When the CR is calculated from inspection
results for a floor that has holed through, the issue becomes when the floor actually holed through. To
account for this uncertainty, the following should be applied:
1) Add 25% to the measured CR if only a couple of small holes (<1/8 in.) were found.
2) Add 50% to the measured CR if one of the few holes were large (>1/4 in.) or if several smaller
holes were found.
3) In case of no leak, the measured CR should be assigned a confidence factor. See section O.4.5.2.
e. Corrosion type (product side) is set to Generalized if not specifically known to be localized from
previous inspection. Note: Soil Side CR is always localized (setting is not optional).
g. Dike – If installed, it is assumed the dike will retain 50% of the Tank content in case of a rupture, i.e.
50% will be released over the Dike.
h. Environmental sensitivity; unless noted otherwise for the specific tank, use a medium setting for
environmental clean-up cost factors.
i. Penalty cost should be applied where relevant, but should typically not be included in the
consequence of failure calculation. This cost is added directly to the consequence cost value.
j. Inspection history; the effectiveness categories should be assigned based on the inspection technique
and efforts documented in the inspection history files. Where a floor and sand pad are replaced in
kind, the previous inspections can be carried forward with the new floor (as relevant knowledge), but
should be ignored in case the circumstances for some reason do change (operational parameters,
different floor thickness, sand pad quality, etc.).
k. Leak detection time should be set relative to the leak detection methods and test frequencies that are
applied at the facility. The following time intervals are suggested as default settings for leak
detection:
m. Material of construction; all tank shell and floor materials are assumed to be carbon steel.
n. Native soil resistivity for concrete pad foundation is set to RPB Installed, which gives an adjustment
factor of 1, i.e. no effect of the calculated CR.
o. Operating temperature; for a given range of operating temperatures, the worst case should be selected
for the analysis.
p. Planned inspection; RBI recommended inspection effectiveness for future inspections is always ‘A’.
This is due to the fact that the increase in actual inspection cost by going from a ‘D’ level to an ‘A’
level inspection is minimal compared to having to re-enter the tank five years earlier for the next
inspection.
q. Product-side corrosion rate – when measured and known, the corrosion type should be set to
Localized in order to avoid potentially counting the measured rate twice. The measured Product Side
CR is expected to a large degree to capture the soil side corrosion as well as the product side
corrosion (i.e. where ultrasonic based inspection techniques are applied for WT measurements).
r. Repair of tank floor; when tank floors are repaired, it should in generally be assumed that the
minimum wall thickness is brought up to minimum 30% of the original WT. This only applies for fill
welding of localized pitted area – preferably the minimum WT has been measured. The WT and Age
can be updated for the analysis, in principle considering the floor as a new item with a reduced WT.
See section O 4.5.
s. Soil-side corrosion rate – which inherently is a Localized Corrosion Type, should be calculated unless
specifically measured.
t. Tank design and maintenance codes; all tanks are normally assumed designed to a recognized
industry standard, and maintained according to API 653 or similar code.
u. Tank height and liquid level; it should be assumed that the average Tank have a liquid height
equivalent to 95% of its total Tank height.
v. Tank Pad; New Sand Cushion (sand pads) are assumed to be Native Soil or Construction grade sand
(depending on whether or not sand test data is available from time of installation). This setting is used
until inspection records show that no localized pitting is expected from the soil side. At this point in
time, the sand pad quality should be upgraded accordingly. This conservatism is to take into account
that even High resistivity, low chloride sand can be contaminated with for example lumps of clay
from the time of installation, and hence cause localized corrosion from the soil side. Tanks with
Concrete Pads are modeled for Soil Resistivity as Tank with RPB.
w. Tank-floor replacement or major repairs; the scenario for the RBI analysis significantly changes in
case the floor is replaced or undergoes major repairs or even just has a Liner added after several years
in service. In these cases, if relevant, the Tank should be analyzed for two scenarios, before and after
the changes have taken place.
x. Wet or dry product – i.e. fluid contains water; generally, it is assumed that all gasoline, diesel, fuel oil
and crude tanks contain a wet product, unless specifically noted otherwise.
b. Consequence calculation; while the CoF category represents a CoF range, and that the CoF is a
product of the Environmental + Financial consequence, one can argue that the Financial consequence
should not necessarily force inspection in the case where the client determines inspection
predominantly driven by regulatory requirements and environmental considerations. Thus, by
omitting the Financial contribution the consequence category potentially becomes lower, allowing a
more optimized utilization of the AST for the specific risk scenario – in this case Environmental Risk.
At the same time, it is understood that the operator accepts the Financial Risk, whatever it might be,
and does not represent a driver for inspection in this context.
c. Cathodic protection; the effectiveness of a CP system highly depends on how it is designed, installed,
monitored and maintained. It can take several years to adjust the system for optimum performance.
Thus, no credit should be given for a CP system, until it is verified by inspection that the system is
indeed effective. The system should be set-up and maintained by qualified personnel and tested in
compliance with an acknowledged standard. A defective CP system can cause extreme rates of wall
thinning, e.g. reversed polarity, defective rectifiers, incorrect voltage or incorrect CP system design
can cause an accelerated rate of wall thinning under circumstances where little or no corrosion would
otherwise be expected. The appearance of this wall depletion would be similar to corrosion of carbon
steel in a moist environment, only at significantly higher rates.
d. External heat sinks for external heat coil can cause an increase in internal shell corrosion at these
locations.
e. Leaks with minimal consequence; for some tanks, where an RPB or concrete / asphalt foundation is
installed, the consequence of a small bottom leak is negligible, and the consequence driver is more the
risk associated to a rupture. In these cases it is appropriate to reassess the inspection target, while
keeping in mind that inspections normally relate to floor thinning and small leaks, and has limited
value for determining degradations leading to rupture.
f. Measured corrosion rate; where a measured CR is derived for a tank floor that is holed through, it is
virtually impossible to determine when hole through actually occurred and as such determine an exact
CR.
g. Moist or wet sand pad; A saturated sand pad (for example by rainwater or deliberate saturation with
soda-ash solution to change pH level), is likely to be more conductive than the native soil beneath.
This will potentially catalyze a high CR. The water is adding oxygen and increasing the O2 level in
the sand pad compared to the native soil. This potentially creates corrosion cell and introduces high
risk of crevice corrosion. It is recommended to maintain the sand pad as dry as possible at all times.
h. Release prevention barrier; the installation of some RPB’s will diminish a CP systems effect. This
should be considered when installing the RPB as well as allowing space for installing CP in between
the tank bottom and the RPB. The underlying assumption for calculating the environmental leak
consequence for tanks with RPB assume that there is a 1/8 in. hole in the RPB and the liquid head is 4
in..
i. Soil resistivity; variation in soil resistivity or sand pad conductivity potentially leads to formation of
macro corrosion cells. This might increase or cause severe soil side corrosion at certain locations
across the tank bottom. It is recommended that Soil Resistivity setting is representative to the Sand
Pad, if possible. In order to consider potential problems in this relation, it is recommended to use
Moderately Corrosive as the default setting for soil resistivity until a high level (A or B) inspection
shows proof of better conditions. This should tie-in to the Tank Pad quality and CP effectiveness
settings, i.e. set these conservatively until proven better by inspection.
j. Stray current; issues related to stray current drainage are addressed under the CP system, and should
be one of the parameters in designing, monitoring and controlling an effective CP system. It is
suggested that testing is done to determine if stray current is a potential problem or not.
k. Tank pad; a clean sand cushion with a high resistivity level and low chloride content will in theory
get contaminated over time by moisture, ionic species, salts and chlorides, etc. from the native soil
beneath. Also, lumps of clay or debris from the floor construction work can contaminate a new tank
pad, and potentially cause localized pitting type corrosion, even with a fairly effective CP system in
place. It is recommended to model a new tank pad fairly conservative (for the CR calculation) until
inspection shows that it is safe to assume that the sand is clean.
l. Leakage without consequence; given the right leak detection system and RPB or concrete / asphalt
foundation, it can for environmentally benign fluids be justified to allow the tank floor to run to
theoretical hole through before inspection, i.e. set the inspection target DF = Max value. This could
be considered, especially given the fairly large degree of conservatism that is built into the
methodology.
b. Tank pad – for sand pads, use native soil if no sand test records are available to document that the
sand quality is high, and use construction grade sand if sand test record is available to document high
quality sand. Use these settings until the pad quality is proven by inspection (a or b level) to be high
quality.
c. CP system – although a CP system is installed and should be working, no credit should give for its
effectiveness until proven by inspection (a or b level).
d. Stored product is assumed wet if in doubt or if the measured cr from previous inspection indicates a
product side CR ≥ 2mpy.
e. Internal tank shell corrosion is determined for each course. Unless inspection data is available, course
1 CR is equal to the highest of calculated and measured bottom product side CR.
f. External tank shell CR is assumed never smaller than 0.5mpy, unless determined specifically through
effective inspections. This can be altered at the user discretion. The main objective with assuming at
least some corrosion is to drive formal inspection recommendations, although corrosion might not be
considered critical.
O.2.5.5.2 Measured Corrosion Rate (CR) and New Wall Thickness (WT)
a. A new floor WT can be calculated based on available inspection results. This new WT is
subsequently adjusted for an intuitive level of confidence in the measured value.
b. Any measured CR is adjusted for a level of Confidence, before applied in the analysis calculations.
c. In addition to the adjustment of the measured CR, the ar/t table is designed conservatively to take into
account the uncertainty about the actual tank floor condition, depending on age and available
inspection that is considered.
d. If the measured CR is derived from data on a floor that has holed through, the following convention is
used to account for the uncertainty of when the initial hole through did occur:
i. In case of only one hole, or a few small holes, multiply the measured CR with 1.25.
ii. In case of several holes or a large hole, multiply the measured CR with 1.5
b. A Liner that is inspected and found in Good condition triggers an adjustment of the Liner age –
maximum 0.6 time actual age (credit given for max 30years actual age). An old liner can never be
considered the same as a new liner, regardless of inspection and repairs.
c. The analysis can be reset to consider a revised wall thickness (see item 2), that is based on available
minimum measured thickness. This procedure requires data from at least one A or B level inspection.
The New wall thickness is adjusted for confidence level before applied in the analysis. Past
inspections are maintained in the analysis, provided everything else remains the same.
d. The ar/t table is designed to be fairly conservative. Hence, the amount of credit given for a Low Level
inspection (D or even C) is almost negligible. The ar/t table (or damage factor – DF – table) is derived
from a statistical concept, outlined in section O.5.7.2 in this document.
b. Native Soil Type or Soil permeability is taken from the average measured value for hydraulic
conductivity (cm/sec). The most appropriate is to use the lowest representative value, which denotes
the smallest soil particle size and compactness that is found in vertical direction of the tank floor
towards groundwater, and which would represent a continuous layer beneath the tank.
c. The fraction of fluid that is calculated to contaminate the soil and groundwater in case of leakage is
based on volume released through a 1/8 hole, distance to groundwater and vertical flow velocity
through the soil. It is not considered how much of the leaked volume is likely to be absorbed to wet
the soil in horizontal direction.
b. Inspection targets are set conservatively. Table 4 gives a suggested set of inspection target values for
the DF. These are developed from a relatively conservative rational. The user is free to set the target
DF to whatever value is found to be appropriate and in line with the company’s inspection and risk
strategies, as long as there is a rational basis for the numbers.
CoF Category A B C D E
Max DF 1390 1390 1390 1390 1390
Limit 30% 24% 18% 12% 6%
Recommended
415 335 250 165 85
Inspection Targets
c. Inspection Credit; In general there should be applied a conservative approach for assessing and
interpreting the available inspection and test records. For example, it is recommended not to take
credit for more than a ‘C’ level inspection for a non-destructive soil side inspection dated prior to
1995, as the typically applied techniques were not adequately developed at this point in time. With
better techniques available today, the focus should be turned towards ensuring that high credit is
achieved for future inspection, as suppose to stretching the credit for past inspections and inspection
practices. This might include special qualification programs for the technicians that apply the NDE
techniques that are commonly used.
concept is to modify a generic failure frequency for the component failures by a factor that is related to
both the potential degradation occurring, and the amount and effectiveness of inspection performed to
reveal the degradation mechanism.
O.3.1.1 Risk
One way to portray risk quantitatively is to produce a point-estimate of risk from the consequence-
frequency data pair. This is usually done by multiplying the likelihood and consequence data points
together to produce a measure with units of consequence per year. The mathematical expression for this
score is as follows:
Equation 1: Risk Calculation
Risk ($/yr) = LoF (events/yr) × CoF ($/event)
Multiplying likelihood and consequence together is convenient because it reduces the risk measure to a
single point. The single risk point is often referred to as the expected value of risk for a scenario, and it
can be thought of as a probability-weighted consequence estimate.
Each type of Risk scenario or event for a given component4, being floor Leak, floor-to-shell Rupture,
constitutes a Risk contribution. Thus, the total risk is the summary of all contributing events. The
calculated total Risk is the expected exposure in relative terms, and does take into account the worst-case
consequence scenarios and probability for each failure scenario. The same applies for the shell
component.
O.3.1.2 Likelihood of Failure
The estimation of a component’s leak frequency is found by using a factor to adjust a base (generic or
typical) failure frequency. This modifier is referred to as the damage factor (DF). The scenarios in the
risk model dictate how to combine failure scenarios for a component.
The likelihood of failure (LoF) is determined as;
Equation 2: LoF Calculation
LoF (events/yr) = Base Failure Frequency (events/yr) × Damage Factor
The Base Failure Frequency is a Generic Failure Frequency (GFF) that has been derived from operator
experience within the petrochemical industry in the US (see Table 2). The damage factor is a reflection of
the condition that the specific Tank component is expected to be in. The damage factor increases as the
related component degradation worsens.
O.3.1.3 Consequence
The consequence of a spill is measured in dollars and consists of environmental clean-up costs,
environmental penalties, repair costs, and lost opportunity costs. The Consequence of Failure (CoF) is
determined as;
Equation 3: CoF Calculation
Total Cost($) = Environmental Clean-up($) + Environmental Penalties($) + Lost Business
Opportunity($) + Repair($)
4
Floor and Shell are considered two separate components in that a Floor might leak at the same time as the Shell
fails. Hence, a Shell failure is an added risk to the Floor failure, and as such not relative to the Floor risk.
The basic approach to estimating the environmental clean-up costs of a scenario is to add the cost for the
various clean-up methods needed to remediate a spill. Each component has a Clean-Up Factor (CUF) that
is based on the location of the spill and the type of material spilled. The unit of measure for the CUF is
dollars per barrel ($/bbl).
In mathematical terms, the cost for each component of the environmental clean-up operation is expressed
as follows:
Equation 4: CUF Caclulation
Environmental Clean-Up Cost = Volume × CUF
The costs for environmental penalties, repair, and lost opportunity are input by the user.
The CoF that determines the consequence category is the probability weighted Consequence (weighted by
GFF for each failure scenario, i.e. for tank bottom this is small leak and rupture).
O.3.1.4 Relative Risk
As a mean to assess and express risk in relative terms, such as a low, medium and high risk, the
qualitative assessments of likelihood and consequence are assigned to categories. A group of risk
categories or squares in a likelihood/consequence matrix make up a Risk Level. As an example, a low
probability is placed in Category 1, and a medium consequence might be assigned Category C. These
values can then be displayed in a matrix.
Figure 4 shows a risk matrix displaying five levels of likelihood and five levels of consequence. Risk
increases from the lower left corner to the upper-right corner of the matrix. So, E-5 would be the highest
risk point on the matrix, and A-1 would be the lowest.
4
LIKELIHOOD
3
Plot of a scenario
2 with a rating of C-1
A B C D E
CONSEQUENCE
Figure 5 provides an example of four risk-levels assigned in a five-by-five matrix. This matrix format is
being used throughout this methodology.
4
LIKELIHOOD
A B C D E
CONSEQUENCE
By assigning levels of risk within a matrix format, the relative risk is easily communicated and decisions
can be made regarding the disposition of the various underlying scenarios. It is important to note that
assigning risk-levels in this way should be a reflection of the company’s policies and tolerance to risk
exposure.
O.3.1.5 Total Risk – Tank Specific
The risk is calculated separately for the floor and the shell components, and can be added up or looked at
separately to make up the total risk associated with the tank (expressed in $/year). However, when
concluding the Risk Level for the tank, the Risk Level for the floor and shell should be considered
separately. Hence, the higher Risk Level determines the tank Risk Level as well as the component that
drives the risk, i.e. floor or shell. See the conceptual approach on Figure 6.
Calculate Tank
Risk Level
Floor Shell
Figure 7: Risk Rank Matrix (LoF and CoF Categories) and Recommend DFTargets
CONSEQUENCE ($)
$10K $100K $1M $10M
1000
LIKELIHOOD (DF)
4 415 335 250 165 85
100
3 - - - - 85
20
2 - - - - -
2
1 - - - - -
A B C D E
CONSEQUENCE (CATEGORY)
RISK LEVELS:
: Low Risk : Medium-High Risk
: Medium Risk : High Risk
Each field of the risk matrix holds the Inspection target for that CoF Category, and relates to the damage
factor (DF) value only. When Planning Inspection for AST’s the objective is to trigger an inspection
when the target is reached, rather than determining the inspection effectiveness necessary to achieve a
target value.
The set of DF target values suggested in the Inspection Planning Matrix relate to a DF range of 1 through
1390, and reflect a certain level of risk tolerance. See section O.8.1.2 for how the target values were
developed. It is recommended that the user evaluates these targets and determines the most appropriate
values that represent the company’s strategy for risk management.
The risk levels (Low, Medium, Medium-High and High) are not specifically used for inspection planning
in this methodology description. It is however recommended to incorporate the Risk Level classifications
in the mitigation strategy that relates to other means of mitigating risk than inspection. For example, the
tolerable risk increase per year (risk will increase with the DF, that is calculated as function of corrosion
rate and time) might be set low for a Medium-High Risk tank compared to a Medium Risk tank. The main
objective is to draw attention to the relatively high risk tanks before they actually are due for inspection.
Figure 8 illustrates the inspection-planning concept. The tank is assumed analyzed well before it is due
for inspection, and a time for the next inspection is recommended. The analysis also shows the credit that
is gained from an inspection that is performed at a given point in time. The inspection can theoretically be
performed before target, at target or after the inspection target is exceeded. The optimum time for the
inspection is obviously at the time the target is reached.
Risk / DF
Inspection Recommended
DF
Target
Inspection
Credit
Predicted Risk
Increase Risk after
Inspection
Time to next inspection
The risk profile is expressed as the DF increases over time. The target is set from the consequence
category only. A tank can be analyzed at any point in time, and the plot can stretch over any length of
time.
Calculate
LoF
Changes in input data:
- New Inspection Records
- Meas. or Est. CR available
- Similar Tank Data
- CP System changed
- Coating or Liner
Any - Repairs (New Age and WT)
changes since - New Floor
original service - etc.
start date? Yes
No Calculate CRFloor
RUPTURE LEAK
No Yes
Liner Installed?
DFar / t
LoFLeak = GFFLeak × × MFLiner
MFGFF , Leak
DFar / t
LoFRupture = GFFRupture × × MFDesign × MFSettlement LoFFloor = LoFLeak + LoFRupture
MFGFF , Rupture
LoFFloor
&
DFFloor = DFar/t x MFLiner
DFFloor
5
If a Liner is installed the Coating effect is set equal to 1, i.e. no Coating assumed installed and effect considered.
Figure 10: Calculation of Corrosion Rate (r) to derive Modification Factor (DFar/t)
Soil Product
Side Side
Bulk Fluid Adjust for Operating Adjust for Water Water Draw
Temperature Temperature Draws (0.7-1.0)
(1.0-1.4)
Table 9: Summary of Conditions for Soil Side Base Corrosion Rate of 5mpy
6
See API RP 651, 1997 Section 5.3.1
A summary of the conditions assumed for the product side base corrosion rate are given in Table 17.
Table 17: Summary of Conditions for Base Product Side Corrosion Rate
* To determine the need for internal bottom lining, see API Recommended Practice 652.
The Coating Age is basically used to indicate whether or not coating is applied at all. If no coating is
applied, the Age is left blank and the adjustment factor for coating quality (Yquality) = 1.0.
It is assumed that none of the typically used coating applications are 100% effective in preventing
corrosion, hence the MFCoating factors do not start at a ‘0’ value.
The following equations are used to calculate MFCoating.
Equation 5: Coating Factor Calculation
[
MFCoating = MIN (YQuality )
FNeed
, FNeed ]
Where YQuality is a function of time and calculated as:
1 1 1
YLow = Age 3 ÷ Age 2 + Age + 0.55
90 30 20
1 1 1
Y Average = Age 3 ÷ Age 2 + Age + 0.4
1200 200 25
1 1 1
Y Above Average = Age 3 ÷ Age 2 + Age + 0.25
2000 170 30
The formulae are derived from plotting practical experience, which is gathered from a population of
approximately 50 tanks containing a range of different products.
The objective with the coating factor is to take the coating film degradation over time into account. The
formulas are empirically derived from the assumption that a coating film can have a 5, 10 or 15year useful
life, which is rated in terms of quality: Low, Average and Above Average.
The coating film may during inspection be found only partly degraded, resulting in repairs as oppose to
full re-application. For these cases it is suggested that a repaired coating film is not considered as good as
a new film, and the coating quality should be reduced one setting downwards from the initial setting.
Note: In case a Fiberglass type liner is applied, the coating factor, MFCoating = 1, regardless whether or not
coating is needed.
As suggested in the calculation of the coating factor, coating that is degraded and is not reapplied,
potentially cause a more severe corrosion rate in the exposed areas than if no coating was applied on the
entire floor.
O.4.4.2.5.2 Fiberglass Liner (FRP Type)
Apart from setting MFCoating = 1, a Fiberglass or FRP type liner does theoretically not influence the
corrosion rate. This needs to be taken into account when setting the Base corrosion rate for Product Side
corrosion. The liner, however, does directly affect the calculated DF (see Figure 9). Also see section O
4.7 for details on the modification factor for liners (MFLiner).
O.4.4.3 Determine combined Tank Floor Corrosion Rate
O.4.4.3.1 Estimate Internal and External Corrosion Rates
The internal and external corrosion rates are estimated by multiplying the base corrosion rate by the
respective adjustment factors. This will produce two separate corrosion rates that are combined as
described below. It is assumed that the soil side corrosion will be localized in nature while the product
side corrosion will be either generalized or localized.
O.4.4.3.2 Combine Corrosion Rates
a. Option 1: If the internal corrosion is generalized in nature, the corrosion areas will likely overlap
such that the bottom thickness is simultaneously reduced by both internal and external influences. In
this case, the internal and external rates are additive.
b. Option 2: For pitting and localized corrosion, the chances are low that internal and external rates can
combine to produce an additive effect on wall loss. In this case, the user chooses the greater of the
two corrosion rates as the governing rate for the proceeding step.
Note: In order to avoid understating the risk, it is recommended that the combined corrosion
rate should not be set lower than 2 mpy.
b. Determine the minimum measured thickness of the tank bottom, WTMin, Measured
d. In case of repairs: It is assumed a corroded tank floor is always restored to 30% of its original
thickness, as a minimum. Thus, when calculating a New WT – with no inspection data, use:
Table 22: Correction Factor (CWT and CCR) for measured Floor thickness readings and CR
7
Factor is based on the assumed minimum level of restoration.
* The methods listed in this table should be applied in accordance with API Standard 653 under the direction of an
API 653 certified inspector.
For example, given three inspections where one of the three inspections is rated ‘A’, while the second and
third inspections are rated ‘B’ and ‘C’ respectively. This would result in a combined number and
effectiveness of 1.75A inspections.
The DF is derived from interpolation between the 1A Column and 2A Column in the ar/t table, using the
corresponding ar/t value.
Note: Maximums of four equivalent inspections in the same category are considered in total.
O.4.6.2 Inspection Planning – Guideline
Inspection Planning is described in more detail in section O.8.
As a general guideline, an RBI recommended inspection is always assumed to be a level ‘A’ effectiveness
inspection. The reason being, that the added cost for an inspection with high effectiveness level,
compared to a less effective inspection, is minimal once the tank has been brought out of service and
cleaned for inspection.
O.4.6.3 Damage Factor (DFar/t) – Leak Failure
To determine the final damage factor for the tank bottom, a dimensionless quantity, known as the ar/t
value is estimated, and a table is consulted to look up the damage factor for the generic failure frequency.
The ar/t is found as follows:
Equation 7: ar/t
ar/t = (age × rate) / thickness,
Where ‘a’ is the age of the equipment, in years; ‘r’ is the maximum corrosion rate in mpy; and ‘t’ is the
original thickness of the tank bottom, in mils. The ar/t method assumes that the corrosion rate r is
constant over the life of the tank. The value, ar/t, is actually the fraction of the original tank bottom that
has been lost due to corrosion.
The calculated ar/t, and the combined number and rating of inspections, are used to determine the damage
factor DFar/t from Table 24.
Note that DFar/t should be interpolated vertically and horizontally in the table. The ar/t table was
developed based on engineering expertise and experience in the field of atmospheric storage tank failures
and corrosion. The theoretical basis in developing the ar/t table is included in section O.5.7.2.
The DF for shell is derived from the same table as the DF for the floor. See section O 5.7 for more details.
ar/t E A B C D
0.05 4 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
0.10 14 1 1 1 1 1 1 1 1 1 1 1 1 3 2 1 1
0.15 32 1 1 1 1 1 1 1 1 2 1 1 1 8 4 2 1
0.20 56 1 1 1 1 2 1 1 1 6 2 1 1 18 10 6 3
0.25 87 3 1 1 1 4 1 1 1 11 4 2 1 32 19 11 7
0.30 125 6 1 1 1 9 3 1 1 21 9 4 2 53 33 21 14
0.35 170 12 1 1 1 16 5 2 1 36 16 8 4 80 54 36 24
0.60 500 133 41 13 4 158 88 49 28 234 158 107 73 346 285 234 192
0.65 587 192 71 27 10 222 136 83 51 309 222 160 115 430 365 309 262
0.70 681 270 119 53 23 305 202 134 89 401 305 232 177 527 459 401 349
0.75 782 370 191 99 51 409 294 211 152 510 409 328 263 635 569 510 456
0.80 890 498 298 179 107 538 416 322 249 638 538 453 382 757 695 638 586
0.85 1005 658 453 312 215 696 578 479 397 789 696 615 543 893 839 789 741
0.90 1126 856 672 527 414 888 787 697 617 963 888 819 756 1044 1002 963 925
0.95 1255 1098 976 867 771 1118 1054 993 936 1163 1118 1075 1033 1209 1186 1163 1140
1.00 1390 1390 1390 1390 1390 1390 1390 1390 1390 1390 1390 1390 1390 1390 1390 1390 1390
A, B, C, and D refers to the effectiveness of inspections. E indicates no credit for inspection. The range of factors is
derived from the generic failure frequency (GFF = 7.2×10-4), such that 1/GFF = 1390, which equates to hole-
through. Note however, that this table was originally developed for the API RBI Software (October 2001), which
for inspection planning only can handle integer type data. Thus, each value has been multiplied by a factor of 10,
and subsequently rounded up to the nearest integer. The GFF correction factor (GFFScale) is taken into account in the
likelihood of failure calculation. The lowest factor of 1 indicates that even under the best conditions, a tank is not
assumed better than 10% of the average population. However, that is not interesting in the context of inspection
planning. A target DF will simulate actual tolerable risk and drive an inspection. Tanks that do not reach the
inspection target level are not subject to further assessment, i.e. the DFar/t is skewed towards higher risk. Hence, the
damage factor should not be summed up for a tank population to express the accumulated risk for a tank farm. This
would result in a exaggerated risk value.
0.80
MF,Liner
0.60
0.40
0.20
0.00
0 5 10 15 20 25 30 35
Liner Age (Yr)
Each of the curves on the graph in Figure 11 are calculated using the following equation, which ensures
that the liner factor ranges from 0.2 to 1.0:
Equation 9: Liner Modification Factor
MFLiner = Min (1, Max (0.2, MFLiner Quality ))
Where the calculated liner factor varies with the quality of the liner, such that for:
a. Low Liner Quality:
1 1
MFLow = × Age 2 − × Age + 0.2
250 70
1 1 1
MFAvgerage = × Age 3 − × Age 2 + × Age − 0.85
1600 48 4
1 1
MFAbove Average = × Age 2 − × Age + 0.2
800 80
1) As default the liner is assumed in Average Condition. The assessment needs to be based on the
liner product, suitability for service, and quality of application work (i.e. experience with the
contractor).
2) Depending on Inspection results, the MFLiner factor is taken from the curve that best corresponds
to the condition of the liner. Note that a Low liner quality can move from Average to Low after
only 4 years in service, whereas a liner can not move from Average to Above Average until ca.
10 – 12 years in service. The maximum expected life for any given liner is 30 years and minimum
15 years (unless the credit factor is entirely eliminated).
b. Make the following adjusts to the liner age based on the inspection results, given that the liner was
found to be in following conditions at the time of restarting service:
i. Good Condition: AgeNew = Age × 0.4 (If Age exceeds 30years, no credit
should be given, i.e. factor forced to 1.0)
ii. Fair Condition: AgeNew = Age × 0.6 (If Age exceeds 30years, no credit
should be given, i.e. factor forced to 1.0)
Example: A liner is installed in 1986, inspection and flood coated during inspection in 1997, at
which point it was considered in good condition. The new liner age (AgeNew) is calculated
as:
It is assumed that for any liner, that all bad areas are repaired before the tank is brought back into service,
in particular for the low quality liners.
8
The base frequency for rapid bottom failures of 2.0×10-5 is based on the experience of one of the large operating
companies. They experienced 9 rapid bottom failures over 40 years for a tank population of 5,000 tanks. The
failure frequency was then halved based on anecdotal evidence (lack of such incidents) at several smaller operating
companies.
Depending on the foundation type, settlement might be more or less likely. Table 26 shows the
adjustment factors for likely settlements:
It is assumed that if a tank settlement is visible, a survey will be carried out, i.e. if no survey is done, there
is a possibility for an edge settlement that is not noticeable. The factors larger than 1.0 are meant to
encourage a survey.
The inspection can be external or internal, and be more appropriate at certain elevations on the shell than
others. In some cases internal inspection of the shell is not absolutely necessary, but could be beneficial,
for example for localized or pitting type internal liquid line corrosion, etc. To a large degree the issue is a
matter of selecting the most appropriate inspection technique and apply it at the most appropriate location
on the tank. Specific inspection issues are addressed under the Inspection Planning section. Although
specific techniques, scope and specifications are considered outside the scope of this project, the
suggested inspection location on the Tanks shell (by course or ring number) is given.
The DF is calculated purely on basis of corrosion damage, and is derived for each shell course, based on
course age, representative corrosion rate, and accumulated credit from past inspections.
A calculation overview is given on Figure 2, which includes floor and shell calculations. Figure 12
illustrates the method of screening the tank shell for brittle fracture and cracking susceptibility, and
calculating a DF for a given corrosion rate. The DF is used for the future inspection planning.
3
No
Start
Maint to
Screening Code 4
Yes
Questions
For each Course:
Welded CR (Est, Calc or Meas)
Brittle Calculate DF Insp. Credit
Fracture Constr. for Course [1, Course Age
Method 6] Course t,nom
=> ar/t Table
1 Susceptible
Riveted
Adjusted DF
Course [1, 6] Risk Ratio
Cracking for Course [1,
based on tmin relations
Mechanisms 6]
Screening See Table 28 for Brittle Fracture and Cracking susceptibility assessment.
Installed – Age (Year) Age or number of years that the equipment has been exposed to the current process
conditions that produced the Measured corrosion rate. The default is the equipment
age. However, if the corrosion rate changed significantly, perhaps as a result of
changes in process conditions, the time period and the thickness should be adjusted
accordingly. The time period will be from the time of the change, and the thickness
will be the minimum wall thickness at the time of the change.
The Age calculation is based on Course 1, and can be repeated for course (2;6), if no
replacements or significant repairs have been done on any of the courses.
Nominal Thickness of Course specific – The actual measured thickness upon being placed in the current
each course - tnom service, or the minimum construction thickness. The thickness used must be the
(inch) thickness at the beginning of the time in service. Nominal thickness is needed for
each shell course.
Weld joints – E Weld joint efficiency (E). As default E is set to 0.7, but can be changed in line with
the design code.
Material Stress Course specific – Material and grade with allowable stress for each course. As
Allowable – S (Psi) default, all courses are assumed to be equivalent to A283 Gr. C material with YS of
30ksi and TS of 55ksi. Allowable stress should be calculated as per API 653.
Fluid Type Representative fluid type or name. This is based on a list of generic fluid groups or
types.
Gravity (G) Fluid Specific gravity input can be entered on three different level of detail.
• API Number
Detection Time (Hrs) Expected time to detect and mitigate a leak failure.
Default is 72 hours
Spill Distribution (%) Based on plot plans and land surveys, estimate how the fluid volume in the event of
a large leak would potentially distribute between the Dike area, Onsite area outside
the dike, Offsite area (outside the premises), and offsite Surface Water.
It is assumed that no or very limited amount of fluid will seep into the ground, and
the groundwater contamination is not likely to occur.
In case a Dike is installed, 100% of the leaked volume is expected to be contained in
the dike area.
CR Index Ratio Use one of the following two options as basis for the consequence calculations:
1. Only Course 1 leakage
2. Distributed probability of leakage over the shell height
Option (1) is somewhat more conservative.
Inspection Target – Manual setting of maximum acceptable damage factor (DF). This is only entered if
DFTarget the automated Inspection Planning option shall be overridden.
Inspection History – Information typically related to the inspection records – see also CR Calc.
Last Insp. Year Course specific – Year that the last inspection was done, and during which fairly
reliable tmm were obtained.
Use last inspection year for Course 1, if information is not available for the other
courses.
tmm Course specific – Minimum measured wall thickness (during the last inspection
year) for each shell course. If not available, tnom will be used. Note that Measured
CR can not be used unless tmm is available.
Inspection Course specific – The rating category of each inspection (internal and external) that
Effectiveness has been performed on the equipment during the time period (specified above).
Separate evaluations are required for internal and external shell CR, but need to
relate to the driver of the combined CR for the purpose of the analysis.
Inspection Count Course specific – The number of inspections on each course (both internal and
external) that have been performed, and that correspond to the effectiveness count.
Internal CR (mpy) Measured internal CR for Course 1 only. Assuming inadequate data is available to
determine CR for Course (2;6), the CR for Course 1 is just a input for a generic
distribution of CR across the shell height.
If adequate data is available to determine CR for Course (1;6), these should be used
as Estimated CR.
Internal / External For both internal and external corrosion, indicate whether corrosion would be
Corrosion Type generalized or localized.
Internal Liner or Indicate if a internal liner (FRP type) or coating (epoxy or similar) has been
Coating, Quality, and installed, what year it was applied, and expected quality of application (low/below
Year Installed average, average or above average). The quality setting should be based on
experience with the liner or coating product in the relevant service and application
contractor. The year installed should be based on original installation date, or the last
date of liner / coating rehabilitation.
Shell Internal Base The expected or observed internal corrosion rate on the shell of the tank. Similar to
Corrosion Rate (mpy) the floor assessment, the CR = 5mpy if the product is considered wet (containing
water) and 2mpy if it is dry.
Internal Lining Yes or No. Is a liner or coating needed to protect the tank bottom and shell from the
Needed? corrosive nature of the product?
External CR (mpy) Representative measured external CR for the shell – all courses.
Climate Type of climate is used as a parameter to estimate external corrosion rate of the
(marine/temperate/ari shell, if a measured or estimated CR is not available.
d)
Insulation and Quality Is Insulation installed (Y/N) and what is the current quality of it. Once insulation has
been installed at a given quality level – in terms of water and weather protection – it
is assumed that repairs, rehabilitation or replaced is made in case the insulation
quality significantly degrades.
Estimated CR Course specific – Combined Internal and external corrosion rate (CR) for each
course. This is derived from the inspection records and is entered as the estimated
CR for each course.
The estimated CR are manually assigned, and can be a combination of measured,
calculated and assumed rated. These should be assigned by a subject matter expert.
• Calculated CR (Default)
• Measured CR, or
• Estimated CR
Insp. Planning
Analyze Year Year for which the DF is calculated. This is the same as the RBI Analysis Year. By
calculating the DF for each year in a selected range, the risk profile is expressed as
the DF plot over time (Year)
Baseline WT Select whether to use tnom or tmm as basis for the calculation of remaining wall and
Critical Course numbers.
tmm should not be used unless at least one A level inspection is available to support
the tmm. Default is to use tnom.
HF (ft) Filling Height – Maximum filling height that the operator is certain to stay within.
There should be a control measure available to ensure this maximum filling height.
S (Psi) Allowable stress for each course. Determined from yield and tensile stress for the
Material of Construction – for each course, as per API 653. If the MoC is not
known, default values are assumed equivalent to A-283 Gr. C, i.e. YS = 30ksi and
TS = 55ksi.
E Weld Joint Efficiency factor (E) – if not known, E = 0.7 is used as default.
HC (ft) Course Height – assuming all courses are the same height. If not, than indicate the
average height of each tank segment to consider. Course 1 is at the ground level.
Note that the corresponding nominal wall thickness (tnom) needs to be updated
accordingly.
G Specific Gravity of the stored fluid. If not known, G = 1.0 is used. This is reasonable
for most hydrocarbon fluids, although it should be noted that the Sp. Gr. can be
much higher for other fluids. For example Sulfuric Acid can have a Sp. Gr. of
approximately 1.7.
Age (Years) Use Installed Year for each course as basis for Age. Some of the upper courses
might have been replaced. This should be taken into account in the analysis.
tnom (in) Nominal wall thickness (tnom) for each shell course (or segment)
tmm (in) Minimum Measured wall thickness (tmm) from the last inspection, for each course –
if available.
Inspection Year Year of the last inspection where good information on wall thickness measurements
was obtained – i.e. year where the tmm was recorded.
CR (mpy) Corrosion Rate – If not using the calculated CR, then enter Measured (only Course 1
CR, based on analyzed inspection results) or / and Estimated CR for each course,
and select which set of CR’s to use – Estimated or Measured (or Calculated).
either environmental cracking or material stress related cracking. Combined with material embrittlement
this can lead to shell fracture and severe failures. Material embrittlement and fractures are much more
difficult to determine and mitigate from an inspection perspective than corrosion phenomena’s. Hence, for
the purpose of this module, cracking has been modeled as a random type failure with a couple of defined
indictors for susceptibility, but without time dependence. The RBI output in that respect is a basis for
further analysis or expert evaluation. Exercises such as determination of Leak Before Break (LBB)
criterion is not part of the RBI analysis, but more so an element of the add-on fitness-for-service scope
which should be initiated once the tank is found susceptible to cracking and critical to fracture.
The rupture scenario is assumed to be associated with a brittle fracture type failure, and is also assumed
mitigated in cases where compliance with the Brittle Fraction Considerations of API 653 is ensured. A
rupture failure can be initiated at a locally stressed area (due to wall thinning) or at a crack like flaw.
Material cracking is considered a separate issue, for which another screening process is applied.
The safe filling height, susceptibility of cracking, and the predicted probability of leakage are determined
as output from the RBI analysis. The probability of leakage is the main driver for future inspections,
whereas the safe filling height and cracking susceptibility should initially lead to more detailed analysis,
and possibly recommended inspections. With low cracking susceptibility, it is assumed that leakage will
occur before rupture, both for the wall thinning and material cracking cases.
O.5.3.1 Corrosion Damage – Basic Assumptions
It is generally assumed that the thinning mechanisms will resulted in an average rate of thinning/pitting
over the time period defined in the basic data. The likelihood of failure is estimated by examining the
possibility that the corrosion rate is greater than expected. The likelihood of discovering these higher
rates is determined by the number and type of inspections that have been performed. The more thorough
the inspection, and the greater the number of inspections, the less likely it is that the corrosion rate is
greater than anticipated.
It is assumed that corrosion would eventually result in failure by total consumption of the shell thickness.
The approach also assumes that internal and external corrosion are additive except when both are
localized in nature. For those cases where localized corrosion or pitting are the dominant contributors to
both the internal and external corrosion then the corrosion is not additive.
O 5.4 Screening for Brittle Fracture and Cracking
Brittle fracture and environmental cracking of the tank shell material has not been considered in detail
with the objective to determine a time for inspection. However, under certain circumstances the shell
might be susceptible to these types of failure modes. In order to evaluate the likelihood of these types of
failures, a screening method is applied to ascertain the level of susceptibility to brittle fracture and / or
cracking.
In case a tank shell is found to be susceptible to brittle fracture and / or cracking, the user needs to initiate
a further analysis of the relevant aspects of the driver for the particular failure modes. This is not covered
under the scope of this RBI model.
O.5.4.1 Brittle Fracture
Based on the requirements of API 650 and API 653 (see section 5 specifically), the following questions
are used to determine susceptibility to Brittle Fracture, and thus indicating if further assessment is needed.
The answers should relate to the lower 3 courses of the tank, as brittle fracture is not considered likely for
course 4 and above.
For example; A further assessment may entail a more thorough engineering evaluation of the relevant
aspects, with the outcome that precautionary operations constraints are introduced, etc.
The objective with the screening method is to define a coarse methodology that can be easily applied with
reasonable accuracy to determine the susceptibility for environmental cracking in a tank shell. For the
purpose of this methodology, the cracking susceptibility has not been considered as a function of time.
Environmental cracking is crack growth caused by simultaneous action of corrosion and tensile stress. It
comprises the anodic process of stress-corrosion cracking (SCC), the cathodic process of hydrogen-
assisted cracking (HAC) and liquid metal cracking (LMC). LMC is not considered for the purpose of this
methodology. Environmentally induced hydrogen cracking has been known by many names, including
hydrogen-assisted cracking (HAC), hydrogen-induced cracking (HIC), sulfide stress-cracking (SSC) and
hydrogen stress cracking (HSC).
Given the diversity and circumstances under which environmental cracking occurs; it is unreasonable to
expect any single mechanistic theory to explain environmental cracking in all environment-alloy
combinations. When the theoretical approach in any field is ambiguous, the practicing engineer must use
the empirical approach, turning to the large volume of field experience and laboratory data that has been
generated over the years. From an engineering standpoint, an understanding of the empirical observations
of the boundary conditions within which cracking occurs, can reduce the need for satisfactory theoretical
explanations, in that, a particular cracking mechanism can be controlled by controlling these conditions.
In general environmental cracking occurs when a critical combination of temperature, tensile stress,
metallurgical structure and environmental conditions coincide. For any given alloy-environmental system,
the engineering parameters of concern are as follows:
b. Metallurgical variables (heat treatment, type of structure, cold work, etc.) which render the allowable
susceptible.
c. Environmental boundary conditions for cracking, such as temperature, solution composition, pH,
electrode potential, necessary impurities, etc.
For the purpose of this model, only tanks that are designed to API 12D or API 650 are considered.
The detailed questions in Table 29 will point out the most likely cracking mechanisms that the shell
material is susceptible to, if any at all. The user also has the option to override this screening exercise
altogether based on expert opinion. In order to consider the shell susceptibility to cracking from a certain
mechanism, all the detailed questions associated to that mechanism need to be answered Yes.
The conclusion from the exercise in Table 29 is a list of cracking mechanisms to which the shell material
potentially is susceptible. The follow-up action might be a more detailed assessment and possibly a
fitness-for-service assessment to ensure that the potential risk is appropriately managed.
Table 29: Cracking of Low Alloy Steel and Carbon Steel Tank Shells – Answer Q1 OR Q2.
9
Sample entries have been made for illustration purposes in this column.
10
If Q1 is not answered, all questions for each cracking mechanism should be sought answered.
The Screening table for Cracking is applied such that the user answers all the questions, and for the
mechanisms where the underlying questions are all Yes (with the exception of Fatigue) will be listed as
potential cracking mechanisms which should be addressed or assessed in greater detail. Any Yes under
Fatigue will induce susceptibility for the Fatigue cracking mechanism. The evaluation beyond this
screening process is not considered part of the RBI study, which merely points out the potential areas of
concern.
Failure Mode (Based on welded tank. Maintained to API 653) GFF Relative probability
distribution
Leak due to corrosion 1.0 × 10-4 98.04%
• Small leak (equivalent to a 1/8 in. hole size) 70%
• Medium size leak (equivalent to a 1/2 in. hole size) 25%
• Large leak (equivalent to a 2 in. hole size) 5%
Rupture due to fracture (material embrittlement and cracking) 1.0 × 10-7 1.96%
• Rupture release of total tank volume (failure of lower 3 courses) 100%
Riveted Tanks are assumed to leak about one order of magnitude more often that welded tanks, i.e.
GFFLeak = 1.0×10-3, whereas a rapid failure or rupture of a riveted tank is consider unlikely.
Welded tanks that are NOT maintained to API 653 are considered to be about 40 time more likely to have
a rapid failure than tanks maintained to API 653, i.e. GFFRupture = 4.0×10-6.
The probability of a given leak occurring on course number 1, 2,… or 6 will vary with the shell height, as
indicated on Figure 13. For this model, the distribution is based on the inspection records for ca. 140
tanks at one refinery in the northern the USA. The distribution is only used for the corrosion rate
prediction calculations, and can be overridden by expert estimate for each individual course. Ideally, the
suggested distribution shown as the CR Index Ratio in Table 31 should be assessed by several more cases,
and possibly altered to account for further supporting information that becomes available. The CR Index
Ratio is considered a variable in the calculations for each tank shell.
52'
44'
36'
28'
20'
12'
4'
The initial set of CR Index Ratios in Table 31 has been derived for tanks with 6 courses. Some tanks have
fewer courses, but will experience similar distribution of CR over the shell height due to liquid level
variations. Since this module in principle only considered the shell by number of courses, the CR
distribution becomes coarser with the decrease in number of shell course. This is somewhat reflected in
the CR Index Ratios for tanks with fewer than 6 courses (see Figure 14).
Relative Corrosion Rate per Shell Course CR Index Ratio for tanks with five or less
Based (6 Courses) Course
2.20
2.00
1.80
1.60
1.40
1.20
1.00
1 2 3 4 5 6
11
The Corrosion Rate Index is used for extrapolating the CR from course one to course two and above – the index
is derived from the graph on Figure 13.
12
There are some tanks with 7 courses, but these do not represent the typical tank and as such the 7th course is
covered under course 6 for this model.
A corrosion rate (CR) shall be determined for each shell course, as the CR can vary significantly with the
height of the shell. The rates can be selected by the user as:
b. Calculated; CR is calculated by the RBI model, using Course 1 CR and the CT Index Ratio for
Course (2, 6).
c. Measured; CR as measured during past inspection, based on tmm and tnom – these should be adjusted
by an appropriate confidence factor.
If possible, measured corrosion rates should be used in lieu of the calculated or purely estimated rate, all
depending on the available inspection data and confidence in the inspection results. The CR can be
measured for some courses and estimated or calculated for others.
Once the CR is determined, it is subsequently applied to derive a DF for the tank shells.
O.5.6.1.3 Water
Water content will either way increase the potential corrosion of the shell. Where the specific gravity (G)
is larger than 1, the corrosion area will be at the liquid line, and where G < 1, the corrosion is expected on
the lower section of course 1. Water can always be expected to be present to some degree, regardless of
water draw practices. Water can enter the tank with the stored fluid, condensation in the vapor space or
ingression of rainwater (depends on roof type and condition).
There are obviously several more damage mechanisms and corrosion aspects that need to be considered,
and it is assumed that the final estimates made by the expert takes all relevant aspects into account.
a. Cone roof type tanks have typically a larger vapor space, and hence more generalized corrosion at the
top half of the shell.
b. Potential shell wall corrosion can be split into three main areas of concern:
i. Vapor space corrosion – typically generalized corrosion, with some tendency of pitting,
depending on the stored liquid type. The problem is confined to the shell area from mid tank
height and up to the top, and is more pronounced for cone roof tanks.
ii. Liquid line corrosion – typically minimal or low pitting type corrosion (depending on stored fluid
type). The problem is found in the range of the fluctuating liquid line.
iii. Liquid phase corrosion – typically minimal corrosion, depending on the stored fluid, found in the
tank section that is continuously in liquid.
iv. Lower Heel corrosion – typically fairly high pitting type corrosion driven by the water content at
the bottom 3-6 in. of the tank. Often the corrosion is prevented by liner application
The following guideline can be used for vertical segmenting the tank shell:
c. Rate of filling and emptying storage tanks (i.e. the liquid line motion) has some effect on the
corrosion rate. Figure 5.27 from /7/ the NACE Materials Selection for Petroleum Refineries and
Gathering Facilities handbook strongly indicates this issue being true. See Figure 16.
13
Only if water can be expected in the fluid and specific gravity (G) is ≤ 1. Water is assumed to be present for all
tanks storing fluids with a specific gravity of 1.0 or less.
Figure 16: Changes in Corrosion Rates with Slow and Rapid Liquid Line Motion
It can be assumed that the tank volume does to a certain degree determine the range and frequency of
liquid line motion. The following is a suggested categorization of tanks on diameter and height. See Table
33.
The above suggests that generically Small tanks will have a relative higher corrosion rate at the liquid line
range than Medium and Large tanks. In estimating the corrosion rate, other relevant factors should be
taken into account, such factor could for example be:
c. Due to increased oxygen solubility, corrosion rates are likely to increase with decreased fluid density,
particularly for Light fluids (API 50 or lighter).
d. Shell wall without internal coating most likely have higher corrosion rates at the liquid line range than
internally coated shells.
e. Large vapor space might increase corrosion. Hence, cone roof tanks (large vapor space – varies with
liquid level) are likely to have more corrosion at the top courses than internal floating roof tanks
(small vapor space). Note that it can always be assumed that oxygen enters the tank (via the seal,
small holes in the roof, lack of purging from last entry, or via the fluid stream). It is not realistic to
exclude presence of air in the vapor space.
f. Vapor space corrosion is likely to be generalized in nature, whereas Liquid line, Liquid heel and
water level are all likely to be localized in nature (pitting type corrosion). Particularly, the vapor
space corrosion is bad if H2S is presence, i.e. Air and hydrogen Sulfide.
Typically a tank has up to 6 and sometime 7 courses (rings or stacks). The typical width of the course
plates is 8 feet. For smaller tanks 10ft plates are sometimes used. Each course might have different wall
thickness (WT) since the WT is determined by the fill height. Since the WT is normally decreasing with
increasing shell height location, the time to leakage due to corrosion can be expected to by highest at the
to of the liquid line range. Conversely, the consequence of a leakage located at the top of the tank shell is
relatively small, compared to the lower shell courses, where the membrane stresses are much higher. This
also supports the postulate that the main concern in relation to catastrophic failures (shell rupture) is
confined to the lower 3 courses.
This DF approach is only applicable for small leaks from welded shells. Frequencies for small leaks
from riveted shells and rapid shell failures are as shown in Table 2 and are not modified further.
Figure 17: Flow Chart to determine estimated corrosion rates for Welded Shells
Product
External
Side
Coating Needed?
Coating Age
Insulation Quality Adjust for Internal Coating Quality
and Age Adjust for Insulation Coating. Low
If Liner (FRP type) Average
is installed, set Above Average
factor = 1.0 (applied according
to API 652)
Coating Quality
and Age Adjust for coating
Calculate CR
Calculate Modified External for Course 2 CR Index Ratio
Corrosion Rate
and up
Under normal circumstances and for the purpose of this assessment the base internal corrosion rate is set
to 2 mpy for dry products. However, if significant bottom sediments and/or water are present, the base
corrosion rate is set to 5 mpy. Table 34 shows the suggested base corrosion rates.
Except for adjustment for Water Draw (not applicable), the adjustment factors for internal shell side
corrosion are the same as for the product side of the floor. See section O.4.4.2 for details.
The CR is only calculated for Course 1, and subsequently extrapolated as per the CR Index Ratio (see
Table 31) for Course 2 and up. Any one of these calculated CR can be overridden by an estimated value.
Dependant on whether or not the shell is insulated, the corrosion rate could increase for certain
temperature ranges. Thus, the base corrosion rate is determined accordingly. The quality of the insulation
is taken into account by a separate adjustment factor.
Table 35 presents ranges of bulk fluid temperatures and corresponding corrosion rates for each climate,
with no shell insulation.
Table 35: Base Corrosion Rates (mpy) for External Shell Corrosion – NOT Insulated
Table 36 presents ranges of bulk fluid temperatures and corresponding corrosion rates for each climate,
for insulated tank shells.
Table 36: Base Corrosion Rates (mpy) for External Shell Corrosion – Insulated
Figure 18 shows the locations for the three climate types in the continental United States. Locations with
a marine climate receive more than 40 in. of precipitation per year or have an average chloride
concentration in rainwater of at least 1.0 mg/l. Locations with temperate climates are assumed to receive
20 in. – 40 in. of precipitation per year. Arid climates exist in those areas receiving less than 20 inches of
precipitation per year.
Low 1.00
Average 0.50
Insulation is assumed to be repaired or replaced when it appears damaged or ineffective, and as such does
not degrade significantly over time, in terms of insulating effect. This should be taken into account when
determining the Insulation Quality.
The factor can also be looked up in Table 38. The lower the factor is the more credit is given. A factor of
1 is equal to no credit.
Table 38: Adjustment factors for quality and age of external coating
Note: If the external shell is pitted, no credit should be given for coating the tank.
The coating quality is categorized as Low (good for up to 5 years), Average (good for up to 10 years) and
Above Average (good for up to 15 years). The quality setting should take in to account past experience
with the coating program for the given environment, contractors workmanship, etc.
For the cases where a continual coating program is enforced and the external corrosion rate is kept at a
negligible level, a measured or estimated corrosion rate should be assigned (0.5mpy generalized is
suggested).
CRMeasured
CR Adjusted =
InspConfidence
Where the InspConfidence is the level of confidence in the particular CR that is derived from the past
inspection records.
InspConfidence :
a. 100%: No Adjustment – Two or more A-level inspections have been performed, and inspection
results have been analyzed to derive an accurate CR.
b. 90%: High Confidence – Minimum one A or B level inspection has been made and the results
have been analyzed to derive the most appropriate CR.
c. 70%: Medium Confidence – Minimum one C level inspection has been made, with a simple
review to derive a representative CR.
d. 50%: Low Confidence – Minimum one D level inspection has been made with a simple review
to derive a representative CR.
a. Calculated Internal CR: Only applicable for course 1. The CR Index Ratio is used to determine
the corresponding Internal CR for Course 2 and up. The adjustments are done before comparing the
Internal CR with the External CR.
The higher of the two CR is considered the critical CR and driver for inspection.
The final CR for Internal CR and External CR, for each course, can be based on the calculated, estimated
or measured rates, or be a combination of all.
This will produce two separate corrosion rates that are combined, for each course, as described below.
a. Option 1 – add corrosion rates: If the internal or external corrosion is generalized or widespread in
nature, the corrosion areas will likely overlap such that the wall thickness is simultaneously reduced
by both internal and external influences. In this case, the internal and external rates are added.
b. Option 2 – use highest corrosion rate: For pitting or localized corrosion on both the internal and
external surface, the chances are low that the internal and external rates can combine to produce an
additive effect on wall loss. In this case, use the greater of the two corrosion rates as the governing
corrosion rate.
Note: In order to avoid understating the risk, it is recommended that the combined corrosion rate
should not be set lower than 2 mils per year.
The ar/t table that was originally developed (see section O.5.7.1) is base on a GFF equal to 7.2×10-3,
which equates to a DFar/t range from less than 1 through 139. It was deemed that integers were more
practical to use than decimal numbers, and a new table was made with a DF range of (1;1390), i.e.
numbers were scaled one order of magnitude and rounded up to nearest integer. This would be equivalent
to creating a table based on a GFF one order of magnitude smaller, i.e. 7.2×10-4.
To make a simplistic extrapolation of these four DFar/t risk profiles to consider up to a total of ‘4A’
inspection effectiveness as well as no inspection, the existing curves have been used to model a Beta
Distribution function, which changes with level of inspection credit.
It is assumed that the Beta Function curves are fairly conservative, and that any corrosion driven
degradation profile will fall within the profile of these curves. By using this Beta Function, any range of
DFar/t can be developed and correlated to a specific GFF. The underlying assumption remains that the plot
profile and relation between A, B, C and D inspection effectiveness is the same. Thus, Table 24 which is
based on GFF = 1/1390 contains one reference plot for each level of inspection effectiveness from no
inspection (E) through the maximum inspection credit (4A), i.e. 17 risk profile curves in total. This DFar/t
is derived directly for inspection purposes, and is comparable for any component – shell (courses) or
floor.
However, in order to calculate the LoF that is comparable between any of the DFar/t–adjusted GFF values,
the LoF calculations based on the DFar/t from Table 24 need to be adjusted relative to the GFF = 7.2×10-4,
which is the basis for the DFar/t table
Instead of using different ar/t look-up tables for each component (floor and shell) and failure scenarios
(leak and rupture), a relative scaling factor (MFGFF) has been developed (see Equation 14). The objective
with the formula is to align the LoF’s calculated for different GFF values using the same DFar/t reference.
As mentioned, the DFar/t table has been developed for a certain GFF value (7.2×10-4), e.g. for which case
the adjustment should be approximately 1. The GFF scaling factor is applied to the LoF calculation as
shown in Equation 15.
GFFc × DFar / t
LoFc =
MFGFFc
Where the subscript ‘c’ denotes that the calculation is GFF specific (component or failure scenario)
Note that the MFGFF is used for the Likelihood calculations only, whereas the unadjusted DFar/t is used as
basis for the inspection planning. Relative to Table 24, the calculated DF for the floor is 10 times larger
than actual and the calculated DF for shell is about 7.2 times smaller than actual. This is taken account
for in the LoF calculations.
All mentioned DFar/t refer to Table 24, and their adjustment for the LoF contribution from each
component and failure scenario is calculated as per Equation 14 and Equation 15.
The underlying assumption for comparing the floor and shell DFar/t is that the prior distribution function
for tank leakage is relatively the same for the tank shell and floor (same DF profile over time). This is
practical in terms of inspection planning, thus applying the DF range in Table 24 for both the floor and
shell inspection planning purposes – un-scaled, and handling the scaling in the DF calculation for each
component when the LoF calculation is made.
Each shell Course is in principle considered as separate component, and as the CR may vary significantly
with the vertical location on the shell, this method will drive inspection towards to location (Course
number) that is most likely to leak.
This approach might appear conservative for inspection planning of the shell. However, the lower
probability of failure for the shell compared to the floor is assumed taken somewhat into account by an
expected lower corrosion rate on the shell wall. As mentioned, the DF is a function of the age (a),
corrosion rate (r), nominal wall thickness (t), and the combined credit for past inspections.
In summary; For the shell assessment, the following steps are made for DF calculations:
a. One DF value is derived for each shell course – from Table 24. These are adjusted with a DF
Adjustment Ratio for Course 2 and up (see Equation 16 in section O.5.7.2) – the DF Adjustment
Ratio for Course 1 = 1.0, i.e. no adjustment.
b. For inspection planning, the adjusted DF (by DFAdjustment, i as per Equation 16) for each Course is used
directly for inspection planning purposes. Any mitigation in terms of refined calculation or added
inspections will target to reduce the DF factors that exceed the DFTarget.
c. The LoF is calculated as a sum of the LoF for each Course (see Equation 23), using the relevant
scaling factor.
O.5.7.2 DFar/t Calculation for Course Two and Up – Relative DF Adjustment Ratio
Assessment of the tank shell is primarily related to, or based on, the lowest course (course 1). However,
the relative risk associated with a shell leak is expected to change with the vertical location of the hole,
such that the consequence from a leak on the lowest course will be relatively much higher than the
consequence from a leak on the highest Shell course.
Furthermore, a tank may have a reduced nominal wall thickness for course 2 and above, compared to
course 1. This means that for a given corrosion rate that the ar/t value will approach 1 (which theoretically
equals a hole through) much more rapidly for the upper most courses than for course 1, and hence
recommending inspection too early and at the wrong location on the shell. This is taken into account by
the following adjustment factor to the DF for each course above course 1. This DF Adjustment Ratio
calculation is used for all courses as relative basis for distributing the one GFF for the shell over the
existing number of courses (see section O 5.9).
t min, Course i
3
DFAdjustment , i
= Min1, 0.3 +
t
min, Course 1
The calculated DFAdjustment Ratio is illustrated on Figure 19 relative to the shell height (Course number),
nominal and minimum required wall thicknesses. The risk ratio is used for adjusting the DF for each of
the shell courses – the adjustment is relative to course one. The graphs on the figure are based on a typical
example of a shell with 6 courses and varying wall thickness. The corresponding tmin is also calculated
and plotted. The lower the Risk Ratio factor, the higher level of risk is acceptable – in relative terms to
course number one.
The Risk Ratio curve shows that for the upper courses a relatively higher level of risk of leakage, or
uncertainty about wall thickness, is acceptable. This is based on the presumption that the higher the
location of the leak, inherently the lower the consequence will be.
The underlying assumption for the DFAdjustment is that each course has been correctly design, with a
proportional corrosion allowance and safety margin relative to course one, and that the Inspection Target
(DFTarget) is < 30% of the maximum DF.
1 1
0.8 0.8
Risk Ratio
Inch
0.6 0.6
0.4 0.4
Thus, the higher the DFAdjustment factor, the less tolerable the associated level of risk is.
The Critical Course Calculation in section O 5.8 shall be considered an added criterion for the RBI
planning.
Note: shell rupture is considered unlikely above half the shell height. A shell rupture failure is also
assumed to cause a total release of the tank volume – in every case. As such, rupture LoF can practically
not be related to a specific course in terms of varying LoF from one course to the other. Thus, the
LoFRupture is considered non-related to DF and only changes generically by whether or not the tank is
designed and maintained to a recognized code, i.e. the LoFRupture is set static at 1×10-7 and 4×10-6
respectively.
The objective with the critical course calculation is to ensure that the Safe Filling Height criterion is
specifically addressed for each course, and that the RBI recommended inspection scope is focused
towards the most appropriate shell course(s). The situation is likely to occur where the DF for one course
exceeds the DFTarget, and another course is actually more critical as far as Safe Filling Height is concerned,
or that the Safe Filling Height is exceeded before an inspection is recommended from the RBI analysis.
Based on the input information listed in the Critical Course Calculation in Table 27, the following
assessment is performed (see API 653 for reference).
With a given tmm at the last inspection year, a selected CR and a Maximum Filling Height calculate the
point in time where the Maximum Filling Height becomes smaller than the Safe Filling Height – or the
predicted wall thickness becomes smaller than the required wall thickness. The course, for which this is
first through, is the most Critical Course.
HF < HSafe or
Where.
14
This could be true in case the maximum filling height is actually below the course itself. For example if course 6
is never expected to see liquid, tmin and tCalc could both be about 0.00. This is in reality not acceptable. Also see
section O.5.8.3.1.
SEt Calc
H Safe = +1+ H0
2.6 DG
Where,
S = Min[0.80YS, 0.429TS] for Course 1 & 2, and Min[0.88YS, 0.472TS] for all other
Courses. S is measured in Psi.
H0 = the distance from the ground to the bottom weld on the course (ft).
tCalc = calculated thickness at the year that the Tank is being analyzed (in)
Where,
tStart = starting wall thickness, which can be either tnom (if installation year is used) OR tmm (if
inspection year is used)
∆Year = time in years between the year associated with tStart and the year for which the tank shell
is being analyzed.
(H F − H 0 )DG
t min = 2.6
SE
Where the liquid column used is from the bottom of the course to the maximum filling height.
The absolute minimum wall thickness for any shell course area is 1/16 in.. This is mostly relevant above
the liquid level.
In the event these added requirements are exceed, the tanks shell should be made subject to a more
detailed assessment immediately.
However, the LoF contribution from shell rupture is equal to the generic GFFRupture, and only depends on
compliance with design and maintenance codes. Thus, the LoFShell varies with the LoFShell, Leak only. The
LoFShell, Leak is a production of the shell corrosion rate(s).
The LoFShell, Leak is calculated based on the course that has the highest adjusted DF. The representative
shell DF is calculated as
The Generic Failure Frequency for the shell leak (GFFLeak) needs to be scaled in order to align it with the
GFF used as basis for the DFar/t factor development. Hence, the LoF for a shell is calculated as per
Equation 23.
GFFShell , Leak
LoFShell = × Max (DFCourse i )× DFAdjustment , i + GFFShell , Rupture
MFGFF , Shell Leak
1 A inspection = 2 B inspections
1 B inspection = 2 C inspections
1 C inspection = 2 D inspections
The inspections need to be assigned in relation to the governing corrosion, i.e. this being internal
or external corrosion.
Table 39: Guidelines for Assigning Inspection Ratings – Internal Tank Shell Corrosion
Inspection
Rating Internal Corrosion
Category
a. Intrusive inspection – good visual inspection with pit depth gage
measurements at suspect locations.
A
b. UT scanning follow up on suspect location and as general confirmation of
wall thickness
E No inspection
Table 40: Guidelines for Assigning Inspection Ratings – External Tank Shell Corrosion
Inspection
Rating External Corrosion
Category
a. Insulated – >95% external visual inspection prior to removal of insulation
b. Remove >90% of insulation at suspect locations, OR >90% pulse eddy
current inspection.
A c. Visual inspection of the exposed surface area with follow-up by UT or pit
gauge as required.
a. Non-Insulated - >95% visual inspection of the exposed surface area with
follow-up by UT or pit gauge as required.
a. Insulated – >95% external visual inspection prior to removal of insulation
b. Remove >30% of insulation at suspect locations, OR >30% pulse eddy
current inspection.
B c. Visual inspection of the exposed surface area with follow-up by UT or pit
gauge as required.
a. Non-Insulated - >50% visual inspection of the exposed surface area with
follow-up by UT or pit gauge as required.
a. Insulated – >95% external visual inspection prior to removal of insulation
b. Remove >10% of insulation at suspect locations, OR >10% pulse eddy
current inspection.
C c. Visual inspection of the exposed surface area with follow-up by UT or pit
gauge as required.
a. Non-Insulated - >25% visual inspection of the exposed surface area with
follow-up by UT or pit gauge as required.
a. Insulated – >95% external visual inspection prior to removal of insulation
b. Remove >5% of insulation at suspect locations, OR >5% pulse eddy current
inspection.
D c. Visual inspection of the exposed surface area with follow-up by UT or pit
gauge as required.
a. Non-Insulated - >10% visual inspection of the exposed surface area with
follow-up by UT or pit gauge as required.
a. Insulated – No visual inspection of insulation surface area or removal of
E insulation.
a. Non-Insulated - <5% visual of the exposed surface area
a. Small Leak Frequency: The small leak frequency for a specific tank is obtained by multiplying the base
leak frequency for small (welded) shell leaks (Table 2) by the adjusted DF.
b. Rapid Shell Failure and Small Leaks: The frequencies for rapid shell failures and small leaks for riveted
tanks are as shown in Table 2 and are not influenced by the damage factor.
Where the Environmental Clean-up Cost + Environmental Penalties are considered Environmental
Consequence, and Lost Business and Downtime costs + Repairs costs are considered Financial
Consequence.
The costs for environmental penalties and lost business opportunity costs are input by the user. The
environmental and repair costs can be included or excluded at user discretion.
subsurface soil 4
5
ground water
RBI Fluid Group Fluid Description Additional fluid types that can be included – based on equivalent
grouping
C6-C8 Gasoline Gasoline, Naphtha, Heavy Naphtha, Light Straight Run, Heptane
*EE; HF; PO; EEA; Methanol; Styrene; Aromatics
C9-C12 and C13-C16 Diesel Oil Diesel, Kerosene
C17-C25 Fuel and Crude Oil Jet Fuel, Atmospheric Gas Oil, Typical Crude, Vacuum Column
Top, Light Vacuum Gas Oil
*Acid (Low, Med and High)
C25+ Asphalt Residuum, Heavy Crude, Heavy Vacuum Gas Oil
*EG; EO
*Fluids are set to an equivalent HC fluid with similar viscosity in centipoise at 75°F
In case of other fluids than hydrocarbons, the user can model the fluid with an equivalent setting from
Table 41, and adjust the clean-up cost factor accordingly.
Due to the inherent inaccuracies in calculating costs related to a given leak, the CUF should be considered
as an indexing score as opposed to an absolute cost value. The CUF represents the relative magnitude of
different consequences that would result if a barrel of oil were released into one impact type as compared
to another. The unit of measure is USD per barrel ($/bbl), which allows for a realistic update of the
relative costs, depending on local conditions, costs and environmental issues. The total environmental
clean-up cost is calculated as:
The CUF’s listed in Table 42 have been based on actual operator experience relating cleanup/remediation
of petroleum spillage. The score range (low, medium, high) that may be selected, is taking into account
that site-specific conditions could significantly influence the consequence. Therefore, the ranges are
intended to allow the user to adjust the CUF score up or down based on product type, surrounding
community and local conditions and/or regulatory requirements. For example, spills of diesel might be at
the lower end of the CUF range and spills containing MTBE would be at the higher end.
NOTE: Spillage of other fluids than hydrocarbon liquids should be assessed as a separate issue, and a
revised set of CUF’s should be applied. The same applies for adverse local conditions that fall outside the
default CUF range (low, medium and high).
Step 2: Estimate the volumes that might be released into each impact type.
Figure 21 illustrates how the methods are to be applied. The figure illustrates the relationship between
release types (tank bottom and shell) and potentially affected areas (contained in dike, groundwater, etc.).
Yes
CUF = $1,500 medium
Bottom Reaches
Groundwater?
CUF = $3,000 high
Leak
No
Reaches No
Surface CUF = $100 low
Water?
No No
Contained Travel
CUF = $50
in Dike? Offsite?
Yes
Rapid Floor/
Shell Failure
CUF = $10
Notes: The impact range should take into account the amount of time it takes to clean up the spill and to mitigate the impact. For
example, a spill that is isolated within minutes may still reach groundwater if the spill is not cleaned up. Contained in dike?
means that the spill does not have the possibility or is given the time to migrate to other areas. For example, it does not reach
surface water or ground water before it is cleaned up.
The number of outage days should include time to blind off the tank and preparation for inspection, tank
cleaning, waste or sludge disposal, blasting, actual inspection work and bringing the tank back into
service.
The cost per day should take costs associated to these activities into account as well. These costs will vary
with the tank size, stored fluid, etc. and as such will have differing contribution to the overall risk.
This aspect has only been addressed for tank floors, as the outage time can vary significantly depending
on the scope of repair work.
The cost and duration of tank repair work is dependent on the size and age of the tank, type of foundation,
type of product (crude oil vs. refined product), etc. The contribution to consequence by repair and
replacement cost is determined from Table 43, which contains a set of suggested default settings.
The following is the assumed distribution of repair efforts for the average case where leakage is found:
a. 25% of all potential bottom leaks will cause Minor bottom repairs, which include floor pick-ups (fill
welding pits).
b. 50% of all potential bottom leaks will cause Major bottom repairs, which includes extensive pit
welding, partly replacement of annular ring or replacement of floor plate sections.
For bottom leak or rupture events, the probability of having to add or install a floor liner or
coating is:
a. 5% for liner – if liner was not installed before the leak and coating is not assessed to be needed
b. 80% for coating or liner – if coating is assessed to be needed and no coating or liner was previously
applied
c. 100% for coating or liner – if coating or liner was installed prior to the leak, i.e. there is a 100%
probability of replacement, extensive repair or flood coating.
Environmental
CoF
Bottom Leak
Fluid Type
RPB,
No Double Floor, No Yes
Fluid type Release Volume is
Asphalt or
is Asphalt set = 0
Concrete
pad
Yes No Environmental
Clean-up Cost
Release Rate Release Rate with from Leak
without RPB RPB CoFLeak = 0
(bbl/hr) (bbl/hr)
Vertical Velocity,
i.e. vertical flow
Soil Type into the soil
(ft/day).
Distance to
Groundwater D (ft)
Penalty Cost
Time to reach
Environmental Groundwater
Sensitivity tgr (days)
Environmental
CoF
Bottom Rupture
Tank Volume
(total stored volume) Tank Size and Max
(bbl) Filling Height
(default = 95%)
No Yes
Fluid type
CUF = $10/bbl
is Asphalt
% of Tank Vol
Yes released outside % of Tank Vol
Dike installed
Dike area retained in Dike
(default 50%)
No
Volume Release
=100% of Tank Vol
% of release
Clean-up Cost Factor contaminating
Onsite Soil Onsite Soil Onsite Soil
Onsite Soil
contamination clean-up cost
Environmental contamination
(bbl) ($)
Sensitivity (default CUF=$50/bbl)
% of release
Clean-up Cost Factor contaminating
Offsite Soil Offsite Soil Offsite Soil
Offsite Soil
contamination clean-up cost
contamination
(bbl) ($)
(CUF)
% of release
contaminating
Clean-up Cost Factor Surface Water
Surface Water Surface Water
Surface Water
contamination clean-up cost
contamination
(bbl) ($)
(CUF)
CoFRupture
Financial CoF
Bottom Leak &
Rupture
Probability distribution of
Input Date:
repair events - for cost Calculate Cost for:
- Floor diameter
calculation: - Repair
- Liner / coating
25%: Minor floor repair - Liner / Coating application
- Outage cost per day
50%: Major floor repair - Outage time
- What to include in the
25%: Floor replacement
CoF calculation
- Repair cost table
Course 6
Course 5 1/8” hole (leakage)
Course 4
Course 3 Dike Wall
30ft
Course 2
CP
Course 1
Dike Liner
RPB
Subsurface Soil Sand Pad (min. 4”)
O.6.4.1 Basis for Bottom Leak Rate and Seepage into Soil
The following section is a brief summary of the approach and technology applied to calculate the
consequence of failure for the relevant tank failure scenarios.
O.6.4.1.2 Head
The storage tank is assumed to have a fluid level of 30 feet. This is viewed as an average head since for
many tanks the level can vary between 0 and 60 feet. Since the flow out of the tank into the subsoil is
through small holes, the 30-ft fluid level is assumed to be constant.
Hydraulic conductivity is also impacted by whether the soil is saturated and by the soil capillary capacity.
The effects of soil suction capacity have been researched and can be significant but have been ignored for
this analysis because making general assumptions about soil suction forces would be unreliable.
API Publ. 581 Appendix O - AST R3 Page 100 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
Note: If the user wants to assume that there are several 1/8 in. holes in the RPB then the flow rate can be
scaled up accordingly.
The API code also provides several designs for leak detection that inhibit fluids from penetrating the earth
grade supporting the tank. The analysis here is for those cases where leaks result in fluids entering the
soil beneath the storage tank.
Three-dimensional flow conditions were analyzed using a relationship that was developed for estimating
the flow from leaks in flexible membrane liners into underlying soils. This empirical relationship was
developed by Giroud and Bonaparte (1989) in SI units:
API Publ. 581 Appendix O - AST R3 Page 101 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
Where:
C = adjustment factor for degree of contact with soil: 0.21 for good contact, 1.15 for poor contact
This relationship was used for estimating flow from holes in the base of tanks into the sand pad and the
soil beneath it and for estimating the flow through the liner into the soil beneath it. This flow rate is
limited to the maximum flow rate from the tank into the open atmosphere which is described below.
To determine the portion that reaches groundwater, the analyst must first determine the type of soil below
the tank pad and the vertical distance to groundwater directly below the tank. Knowing the type of soil
under the tank, the analyst goes to Figure 26 and uses an appropriate hydraulic conductivity and hole size
to determine the flow from the tank.
Again knowing the hydraulic conductivity, the downward vertical velocity in feet per day is found from
Figure 27 15 . The time to reach groundwater (days) is calculated by dividing the vertical distance to
groundwater (ft) by the downward vertical velocity (ft/day). (It should be noted that the downward
velocity model, as represented in Figure 27, tends to be conservative as it assumes that there is enough
product being released to wet all of the soil/sand surfaces as the spill travels downward).
Knowing the time to reach groundwater, the total leak time is used to determine if the groundwater is
affected and, if so, how much hydrocarbon fluid that has accumulated in the groundwater. If the duration
of the leak is shorter than the time to reach groundwater, then the spill is assumed not to affect
groundwater. The volume that leaks from the tank is found by multiplying the leak rate (Figure 26) by
the duration of the leak. The user can utilise their own estimate of leak duration or use the values shown
in Table 47.
15
If there are multiple soil layers then the characteristics (thickness and hydraulic conductivity) of the most
impermeable layer may be used; particularly if this is also the thickest layer. In other cases this may be inaccurate
and a more detailed analysis may be required.
API Publ. 581 Appendix O - AST R3 Page 102 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
If the spill can reach groundwater, the volume that affects groundwater is found by estimating the total
spill using the leak rate and leak duration time, and then multiplying by the Fraction Contaminating
Groundwater as calculated in Table 46. The remainder of the total spill volume would be assumed to
contaminate the subsurface soil.
For a more simplified approach to the release rate from a tank and its downward velocity, Table 45 and
Table 46 can be used instead of Figure 26 and Figure 27.
API Publ. 581 Appendix O - AST R3 Page 103 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
1.E+00
1.E-01 Head = 30 Ft
1.E-02
1.E-03 Gasoline
Diesel Oil,
1.E-04 Light Fuel Oil
Crude Oil,
1.E-05 Heavy Fuel Oil
1.E-06
1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02
Hydraulic Conductivity of Soil Beneath Tank Pad, cm/sec
Table 45: Release Rates for Small Bottom Leak (bbl/hr) – Without RPB and ≤0.125 in. hole
API Publ. 581 Appendix O - AST R3 Page 104 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
1.E+03 Clay Sandy Clay Silt Very Fine Sand Fine Sand
1.E+02
1.E+01
1.E+00 Gasoline
1.E-02
Crude Oil,
Heavy Fuel Oil
1.E-03
1.E-04
1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02
Hydraulic Conductivity of Soil Beneath Tank Pad, cm/sec
API Publ. 581 Appendix O - AST R3 Page 105 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
1.E-02
Head = 4 Inches
Hole Diameter = 1/8 inch
1.E-03
1.E-04
Gasoline
1.E-05
Diesel Oil,
Light Fuel Oil
Crude Oil,
Heavy Fuel Oil
1.E-06
1.E-07
1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02
Hydraulic Conductivity of Soil Beneath Tank Pad, cm/sec
Table 48: Release Rates from Small Bottom Leak (bbl/hr) - With RPB and one 1/8 in. hole
API Publ. 581 Appendix O - AST R3 Page 106 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
∆Hi
P=atm
H0,i = H-∆Hi ; Course i
P=ρg∆H
Rr
H0 =0; Course 1
The leak scenarios and leak rates shown in Table 49 are based on 95% filling height of a 48ft shell, with
each course height being 8ft. The leak rates are calculated at the bottom of each shell course.
API Publ. 581 Appendix O - AST R3 Page 107 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
Table 49: Calculated Leak Release Rate for Flow to Atmosphere (for all fluid types)
The probability distribution for hole sizes over the tank wall height or courses is based on actual
measured corrosion rates at a site with approximately 140 tanks. The trend of this information is
supported by reference /7/, Dr. Richard A. White, Materials Selection for Petroleum Refineries and
Gathering Facilities, NACE International, 1998, Houston, Texas (see Figure 16).
The volumetric release rates (Rr) are determined as function of ∆H (liquid height above the discharge
hole) – assuming the hole occurs at the lower weld on each course.
πd 2
Rr = C d 2 g∆H × 4.45
4
Where,
Rr : Volumetric Flow Rate (bbl / hour), calculates a jet velocity × cross sectional
area of the hole.
Cd = 0.61 : Discharge coefficient (calculated as Cv × Cc) takes into account fluid friction
and jet stream contraction at the release orifice.
Cv relates to the friction at the edge of the hole. The velocity (Vjet) without friction loss: Vjet
= (2g∆h)½. The actual velocity (Vact) is calculated a Cv×Vact, where Cv is set between 0.90
and 0.99. For a fairly sharp edged hole which can be expected in the case of a tank shell leak,
the factor Cv is set to 0.98.
API Publ. 581 Appendix O - AST R3 Page 108 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
The fluid velocity will reduce the flow of mechanical energy, converting it into heat within the
boundary layers hugging the spout wall. Hence, the fluid jet beyond the wall will have a slightly
smaller diameter than that of the spout. The cross sectional area used for the release rate
calculation is this smaller area of the jet stream a short distance from the orifice (sometime
referred to as vena contracta). The difference or ratio between the orifice area and vena contracta
is taken into account with the Cc value, such that AVena Contracta = Cc × AOrifice. The discharge
contraction coefficient (Cv) depends on the Reynolds number, but is typically set at 0.62 for most
fluids.
1 bbl oil = 5.61 ft3 and 1 ft3/sec leak = 15388.5 bbl/day = 641.19 bbl/hr. 1 ft2 = 144 in2.
Hence, 641.19/144 = 4.4527.
In case the filling height is different than 95%, this becomes a significant variable. The lower the level
becomes, the lower the release rate can be assumed.
If the leak will be contained in a dike, the volume of the spill is simply multiplied by the clean-up cost
factor (CUF) to calculate the cost.
Based on the design of the dike (or lack thereof) and the estimated volume released, determine if surface
water or offsite clean up may be a major concern. If the spill can reach surface water, multiply the total
volume of the spill by the appropriate CUF to determine the cost.
If a spill does not reach surface water and is detected quickly, it is assumed that no seepage into the
ground occurs and that a surface clean up is sufficient. If a surface cleanup is not sufficient, the
consequence models further discern whether or not the spill can go offsite. If it can travel offsite, higher
CUFs are used in the cost calculation.
It can in most cases be assumed that the leak duration will not occur unnoticed long enough for the fluid
to seep into the ground and contaminate sub-surface soil or groundwater.
API Publ. 581 Appendix O - AST R3 Page 109 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
courses. The weighted probability of leak is assumed equal to the relative average corrosion rate for each
course. Hence, the consequence model does in reality ignore the driving mechanism for corrosion, and is
at the same time not considering that typically the lower courses have thicker wall than the upper courses
– i.e. this being modeled relatively more conservative for the lower and more critical courses. One could
say that the lower courses have a higher minimum required wall thickness, and as such balancing out the
conservatism.
The theoretical leak location on each course is assumed at the lower course weld seam with the associated
liquid head. For the default scenario, the values in Table 50 would be the total release volumes. The
values in Table 50 can be used as a generic scenario only adjusted for Leak Detection Time and actual
tanks shell height.
The representative release rate (Rr) weighted by hole size for each shell course is calculated as per
Equation 30 as function of hole diameter (d) and course/segment liquid height, ∆Hi (see Table 49). The
CoFLeak for the shell is weighted on the release volume calculation for each course, using the CR Ration in
Table 31.
API Publ. 581 Appendix O - AST R3 Page 110 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
The release rate calculation can be modeled with any number of segments, regardless of the actual
number of courses on the tank. There is no restriction on tank or individual course/segment height as
such, but given that the liquid head (∆H) determines the leak rates, all calculations are geared towards that
and need to be adjusted accordingly if input parameters change. In that case, Equation 31 is applied in the
release rate calculation for liquid height for each of the segments.
Where,
∆H’(n) = Liquid Head (ft) for the lowest line of each segment – for n segments
The pre-assigned leak probability distribution or Index Ratio mentioned in Table 31 is applied for 6
segments, and needs to be adjusted in case the number of segments / courses changes.
In case adequate information and knowledge is available to determine a different CR Index Ratio, this can
be applied and the consequence calculations (based on release volumes) will reflect this change.
The consequences of the instantaneous loss of the tank volume will result in a wave of product, most
likely partly overflowing the dike wall. This type of release is typically caused by a brittle fracture failure
similar to the Ashland tank failure in Floreffe, Pennsylvania in 1988. The following analysis assumes that
the dike walls are not washed out by the release.
A storage tank is often surrounded by a dike that is designed such that the volume of the dike will hold
110% of the volume of the biggest tank in the dike. However, there are theoretical and experimental
evidence to suggest that this design strategy will not completely contain the spill resulting from a rapid
release of the tank volume. An attempt has been made to develop a model that will provide a rough
estimate of the amount of liquid that overflows the dike. The model is based on predictions made using
computational fluid dynamics.
Table 51 presents the fraction of the tank contents that would be expected to overflow the dike subsequent
to a rapid shell or floor-to-shell failure.
0.4 0.05
API Publ. 581 Appendix O - AST R3 Page 111 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
0.5 0.2
0.6 0.35
0.7 0.5
0.8 0.6
0.9 0.7
For practical purposes a default value of 50% of the total tank volume is suggested to overflow the dike.
Thus, the cost of an instantaneous release will be 50% times the CUF for dike clean-up and 50%
distributed to areas outside the dike.
The available GFF’s are used to weigh the CoF between leak and rupture. However, for a tank shell with
several courses and one combined GFF for leakage, the corrosion rate Index Ratio (see Figure 13 and
Table 31) is used to relatively proportion the CoF contribution from each course to the overall CoFShell,
Leak.
GFFLeak GFFRupture
CoFFloor = × ∑ CoF ' Leak + × ∑ CoF ' Rupture
GFFLeak + GFFRupture GFFLeak + GFFRupture
The CoFFloor is a fixed number and a characteristic for the specific tank. This CoFFloor is used as the floor
contribution for determining the Consequence Category for the Tank, and as such largely discriminates
the High risk tanks from the Low risk Tanks. The floor consequence category is used to determine the
DFTarget for the floor.
The Consequence Category is determined from the Consequence range on the Risk Matrix (see Figure 7).
API Publ. 581 Appendix O - AST R3 Page 112 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
The CoFShell, Leak is expressed in terms of Clean-Up Cost Factors (CUF), exactly the same as for the floor
failures. The CoFShell, Leak is a function of the volume released (VolReleased) and how this volume is likely to
distribute over the area around the tank, e.g. varying clean-up cost factors (CUF).
The total VolReleased is weighted on the release volume calculation for each course, using the CR Ration in
Table 31. The total volume release in case of a leak is calculated as per Equation 33.
i =1
Where,
Rtime = Release time, or time that is expected before the release is mitigated.
Where,
CUFLocation, j = Clean-up cost factor (CUF) for Location j. The four Locations around the tank
are: I = Onsite; II = Offsite; III = Surface Water; IV = Dike Area.
API Publ. 581 Appendix O - AST R3 Page 113 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
Please note that in case a Dike is installed, a Leak is assumed not to reach beyond the Dike area, i.e.
CoFShell, Leak = VolReleased×CUFDike Area.
Where,
VolTank = Total tank liquid volume that will be released in case of a rupture
The GFF is a factor of 40 larger for a tank not maintained to API 653.
API Publ. 581 Appendix O - AST R3 Page 114 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
The contributors to the CoF value are ultimately defined by the user, and may include CoFFloor or CoFShell,
or both, and the CoFFloor may include Environmental CoF or additionally Financial CoF. However, it is
recommended that the floor & shell are assessed separately, such that only the relevant aspects will drive
the RBI recommended inspections.
API Publ. 581 Appendix O - AST R3 Page 115 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
A. Determine CoF with DTTargets for shell and floor (separate target values).
B. Determine DF for shell and floor, and derive RBI recommended year for inspection.
Figure 30 illustrates how the CoF Category is derived to establish an inspection target (DFTarget).
API Publ. 581 Appendix O - AST R3 Page 116 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
FLOOR
SHELL
Inspection Target
DFTarget
CoFF Category CoFS Category
Separate for Shell
and Floor
API Publ. 581 Appendix O - AST R3 Page 117 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
DFar/t, S, L
DFar/t, F, L
Course (1,6)
Recommended Recommended
Floor Inspection Inspection Year is
Year when the max Safe Filling Height
adjusted DFar/t, F, L compared to
reaches DFTarget Operational Filling
OR Height
Operating fill height is
critical
Determine
Inspection time Recommended
and scope Shell Course
Inspection Year
API Publ. 581 Appendix O - AST R3 Page 118 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
(Equation 23)
API Publ. 581 Appendix O - AST R3 Page 119 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
Risk ($ / yr ) = LoFLeak ( Env. CoFLeak + Fin. CoFLeak ) + LoFRupt ( Env. CoFRupt + Fin. CoFRupt )
The user can include either the Environmental or the Financial risk contribution, or both in the calculation
for the tank floor.
API Publ. 581 Appendix O - AST R3 Page 120 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
The shell rupture risk does not relate directly to the DF. The LoF part of the risk of shell rupture is equal
to the GFFShell, Rupture. The clean-up cost factor (CUF) has been derived for the total CoF calculation and
the proportional contribution is determined by GFFRelative Ration in Table 2.
( Yr ) = GFF
Risk Rupture $ Re lative Ratio (GFF Rupture × CUF × VolTank )
The risk associated with a rupture is not related to a DFar/t or other adjustment factors. However, it is
assumed that the screening for brittle fracture and cracking both indicate low susceptibility.
( Yr ) = Risk
Risk Shell $ Leak + Risk Rupture
Note that riveted tanks are assumed to have no contribution from the Rupture scenario.
API Publ. 581 Appendix O - AST R3 Page 121 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
The principle approach for setting the Target DFar/t values is using a fixed inspection target factor
(DFTarget). The Fixed Inspection Target Factor determines an all-time maximum limit for the acceptable
level of risk, and is set for each consequence category.
The following maximum Inspection Target setting for each Consequence Category is suggested.
A B C D E
CONSEQUENCE
The target factors in Figure 34 are derived from the background material used to develop the ar/t table
(Table 24). The following aspects should be considered in relation to setting these factors:
a. The DFTarget is used solely to trigger an inspection, not to determine the level of Risk that is to be met
or effectiveness of recommended inspections. The time is a variable for the AST inspection planning,
whereas the time is fixed for planning inspection of most other equipment types (Turnaround based).
b. The recommended inspection effectiveness for AST’s is normally A or B, not lower. This is given
that the added cost for a high level inspection versus a lower level is minimal once the tank is opened
and prepared for inspection. Thus, the DFTarget is not suppose to drive the inspection Effectiveness,
just the timing.
c. The DFTarget addresses an ar/t range from 0 to 1 (where 1 is theoretically equal to through wall hole).
This increased WT fraction compared to pressurized equipment justifies a relatively higher target for
the DF for AST’s.
API Publ. 581 Appendix O - AST R3 Page 122 of 123 June 2003
API Publication 581 – Risk-Based Inspection Base Resource Document
Appendix O – Aboveground Storage Tanks
Optionally, the target factors can be set according to the level of risk that a company decides to apply as
part of their operating strategy. Four risk levels are developed and represented on Figure 36. The
suggested option is Average, and equates to the numbers in Figure 35.
Figure 36: Optional Levels for setting Target DF as function of CoF Category
800
600
400
200
0
0 1 2 3 4 5 6
-o0o-
API Publ. 581 Appendix O - AST R3 Page 123 of 123 June 2003