Risk Based Inspection PrioritizationApplied to an Ammonia Plant

A risk based inspection was performed in an ammonia plant. Risk based inspection is a subset of arisk based management system. It gives management the tools it needs to make cost I benefit

decisions regarding inspection and inspection related maintenance activities.

Lily Sweet, Lynne C. Kaley, Ricardo R. Valbuena and Alan WarnockDet Norske Veritas (U.S.A.), Inc. Houstin, TX 77084

Introduction

Following successes in applying a risk basedapproach to equipment management in therefining and petrochemical industry, a risk

based inspection (RBI) assessment was performed inan ammonia plant. Ammonia plants experience damagemechanisms such as high temperature hydrogen attackin the reforming section and shift conversion sections,amine stress corrosion cracking, and CO2 thinning inthe CO2 absorption/regeneration section and externalcorrosion under insulation. These damage mechanismsalong with others such as brittle fracture were used todetermine the likelihood of failure for equipment items.This information combined with a consequence analy-sis was used to quantify risk for each piece of equip-ment. Risk prioritization enabled the plant to deter-mine the optimum level of inspection or inspectioneffectiveness required to maintain the risk of equip-ment at its present level as the equipment ages, or toreduce the risk of equipment if the present level of riskis unacceptable. The ammonia plant assessment wassuccessful in that it confirmed the plant's major areasof concern and confirmed that the inspections per-formed in the past were effective. Some equipmentitems were identified as high risk by the assessment

and will require additional inspection during the nextturnaround. Other items were identified as low risk.As a result, the inspection scope on these items will bereduced or their inspection intervals will be extended.The plant will be able to increase the safety of the unitby minimizing the risk through effective use of itsinspection funds. The inspection and associated main-tenance cost savings were estimated at $237,500 for areturn on the investment (ROI) of 5:1. The annualizedsavings were estimated at $47,500, which equated to a5% reduction of their maintenance and inspectionbudget.

This article describes a project that was conducted inan ammonia plant. The goal of the project was toincrease the safety of the plant by minimizing the risk.RBI was used to achieve this objective by optimizingthe inspection plans. By evaluating the likelihood of afailure and the potential consequences (that is, therisks) in the fixed equipment and associated piping,inspection plans were developed to minimize operatingrisks. RBI offers a tool to evaluate the present risk of aplant based on past inspection history, process operat-ing conditions and fluids, and damage mechanismsactive in the plant. The present risk can then be forecastinto the future and inspection, new metallurgy, newdesign, or other mitigation efforts can be directed to the

AMMONIA TECHNICAL MANUAL 147 2001

equipment items, which present the greatest risk.The RBI methodology has been in development

since the early 1990s because companies in the refiningand petrochemical industry needed risk-based prioriti-zation and management systems. The AmericanPetroleum Institute (API) RBI methodology, used inthis study, was developed as an industry-sponsoredproject under the direction of the API. This methodol-ogy has gained acceptance in the refining industry dur-ing the past three years and, more recently, in other sec-tors of the petrochemical and chemical industries.

Methodology

In general, the RBI methodology can be qualitativeor quantitative. The qualitative methodology is gener-ally used to quickly prioritize various operating units orsections of plants, based on risk and to identify the highrisk systems. A more detailed analysis using a quanti-tative methodology can be applied to the higher riskunits or sections of plants. In this way time and effortare directed to gathering data for higher risk items. Thequantitative method prioritizes equipment, typicallyincluding piping, by calculating likelihood and conse-quence values for every piece of equipment in the sys-tem.

In this study quantitative method was applied. Thesystems included in the prioritization were the desulfu-rization reactors, primary and secondary reformers, COshirt converters, CO2 absorption, methanator, and NH3

synthesis. Fixed equipment and piping were includedin the analysis. The project began with the collection ofprocess, equipment, and other information from theplant management database, PFDs, P&IDs, materialbalances, and inspection records. Each equipment itemwithin the plant was evaluated with regard to likeli-hood of failure, consequence of failure, and remaininglife. The consequence and likelihood for each scenariowere combined to obtain the risk. Both current andfuture risks were estimated and used to develop the riskprioritization.

Likelihood of failure analysis

The likelihood of failure is dependent on the envi-ronment/material interaction, actual design data, oper-

ational history, experience with this or similar services,and so on. Figure 1 illustrates the different elements ofthis analysis. During this phase of the analysis, poten-tial damage mechanisms are identified and past inspec-tion histories are gathered to identify active damagemechanisms and measured damage rates. The analysisis done using the systematic approach and methodolo-gy developed, as outlined in the API base resource doc-ument on risk-based inspection (API 581) APICommittee (1998). The information is analyzed usingdifferent "technical modules," which are a systematicmethod used to assess the effect of specific failuremechanisms on the likelihood of failure. The followingdamage mechanisms are available in the methodology.

• External corrosion, that is, general or localizedatmospheric corrosion, corrosion under insulation(GUI), and so on, and external cracking for austeniticstainless steels.

• Internal corrosion, that is, general and localizedcorrosion in various environments (hydrocarbons con-taining water, sea water, water-injection systems,amine treating, CO2, hydrochloric-, sulfuric-, and

hydrofluoric acids).• Stress corrosion cracking in various environments

(caustic, amine, chlorides, H2S, and so on).

• High-temperature phenomena (oxidation, hydrogenattack, thermal fatigue).

• Fatigue caused by vibration and flow effects (slug-ging/choking).

• Brittle fracture (low temperature/low toughnessfracture, temper embrittlement, 885 degree embrittle-ment, and sigma phase embrittlement).

The damage mechanisms of particular concern in theammonia plant study included external corrosion, cor-rosion under insulation, high-temperature oxidation,high-temperature hydrogen attack, carbonic acid thin-ning, and amine cracking. Each of these damage mech-anisms is handled by a technical module presently inthe software.

The calculation method for likelihood of failure asso-ciated with the failure mechanisms is based on struc-tural reliability analysis (SRA) Madsen et al. (1986). Inthis approach the stochastic uncertainty in the basicvariables, in particular the uncertainty in the determi-nation of the damage rate and the inspection effective-ness, are taken into account. An important feature of

AMMONIA TECHNICAL MANUAL 148 2001

DegradationMechanism

_jk Damage Loads v.•ff W Strength

N.Da mag e Rate

Model

Mitigtion

L>

4mr

Failure _Mode •

\Limit State

jk Conséquence^

ijk Probability"̂ forfaiture

Figure 1. Elements in the probability of failure modeling.

Failure Frequency Site Specific Modifiers

EquipmentType

PumpColumnTanksPiping

Hole size1/4" 1"4" rupture

X Equipment Factor X Management System Factor

Technical subfactor

• Damage rate

-Inspecton effectiveness 100

Universal subfactor 10

— Plant conditionCold weather

— Scienic activity

{Mechanical subfactor

—Equipment complexity—Construction code—Life cycle—Safety factors

Process subfactor I

TUO

Management System Evaluation Score% (vs. Reqs. of API RP 750"Management of Process Hazards" andother API codes).

—Continuity— Stability

—Relief valves

Figure 2. Calculation of the likelihood factor.

AMMONIA TECHNICAL MANUAL 149 2001

this theory is its ability to include both the prior dam-age estimates and the outcome from the inspections inthe derivation of the updated posterior likelihood offailure (Bayes' Theorem).

The level of information gained from an inspectiondepends heavily on the quality and extent (coverage) ofthe inspection carried out. The inspection quality ismodeled either by discrete probabilities for the inspec-tion effectiveness for the inspection method applied.The actual likelihood of failure analysis is a complexcalculation which uses the generic failure frequency ofeach equipment type as a starting point for the proba-bility ananlysis. The generic failure frequency is thenmodified by factors such as the equipment factor andthe management system factor. The components ofthese factors are shown in Figure 2.

Consequence Analysis

A simplified consequence modeling methodology isapplied, similar to that used in a traditional quantitativerisk assessment. The methodology is based on anevent tree approach and uses pre-simulated effect-sce-narios to simplify the calculational time required foranalysis. The following factors are considered in theconsequence calculation: fluid type and phase, toxicity,inventory available for release and type of leak(leak/rupture). The consequence effects are reported asfour main consequence categories, as follows:

• Flammable events (fire/explosion resulting in per-

Likelihood Consequence CategoryCategwy A B C D E

128

13

11

100

48 16 41 91 59

Figure 3. Risk rankingmatrix with typical

equipment distributionfor the quantitative

method.

sonnel injury and equipment damage.• Toxic releases (personnel injury).• Environmental risks (cost of environmental clean-

up).• Business interruption (lost/deferred production)

and asset repair after failure.The consequence of flammable and toxic events are

Table 1. Adjustments to Release Magnitudes for Mitigation Systems

Mitigation System

Inventory blowdown, coupled with isolation or shut-down systems activated by operators in the control roomor directly from process instrumentation or detectors,with no operator intervention.

Fire water deluge system and monitors.

Fire water monitors only.

Foam spray system.

Consequence Adjustment

Reduce release rate or mass by 25%.

Reduce consequence area by 20%.

Reduce consequence area by 5%.

Reduce consequence area by 15%.

AMMONIA TECHNICAL MANUAL 150 2001

typically used to calculate risk based on personnel safe-ty and equipment damage. Another method of measur-ing consequence is financial. Cost data related to lostproduction, outage, asset repair, adjacent repair andenvironmental cleanup are used to calculate financialrisks.

Risk Ranking

Risk is a function of likelihood and consequence offailure. In a RBI analysis, each of consequence andlikelihood of failure are categorized in 5 groups givinga total of 25 risk combinations. A risk matrix is aneffective way of representing the risk combinations.Each piece of equipment can be placed on the matrix.It allows comparison of plant risk at a given point intime and helps prioritize the risk reduction efforts fordifferent levels of risk (Figure 3).

The consequence of failure was reported in two maincategories: one related to personnel safety and onerelated to economic losses. A target level of acceptablerisk, or risk criteria, was determined. Each individualcorporation typically defines acceptable levels of risk.DNV can provide guidance to corporations interestedin developing its individual level of acceptable risk for

use in RBI and other safety management areas. Theacceptable level of risk was used to determine anacceptance line in the risk matrix. For componentsabove the acceptance line, steps were taken to reducethe risk. For components below the acceptance line,other options exist such as using the financial risk tooptimize the risk reduction effort. This risk matrix wasused to direct inspection/maintenance planning asdescribed in the next section. Performing inspectionreduces likelihood of failure. Reduction of the conse-quence can be achieved by improving detection andmitigation systems. The API RBI analysis has the capa-bility to take into account detection and mitigation sys-tems that would reduce the amount of material releasedduring a loss of containment. The credits given for var-ious types of mitigation systems are shown in Table 1.Similar credits are given based on the type of detectionand isolation systems shown by the plant (APICommittee, 1998).

Risk Based Inspection Planning

Risk based inspection (RBI) planning involvesfocusing the inspection efforts in order to reduce risk offailure. Therefore, an essential part of RBI planning is

Table 2. Risk Criteria Used at this Ammonia Plant

Consequence Category

E

D

C

A,B

Damage Factor

>=100>=10and <100

<10

>=1 ,000>=10and<1,000

<10

>=100>=10and <100

<10

>=1 ,000>=10and <1,000

<10

Suggested Damage Factor Reduction

25X10X

2X

10X5X1X

5X2X1X

3X2X1X

AMMONIA TECHNICAL MANUAL 151 2001

to establish the most cost-effective approach of satisfy-ing the failure acceptance or acceptable likelihood offailure criteria. The key to RBI planning is to use themethod of probabilistic inspection updating, as a cen-tral part of the RBI concept. The methodology to estab-lish the inspection interval is based on selected combi-nations of inspection methods, that is, inspection effec-tiveness, number of inspections, and inspection inter-vals that can ensure that the risk (area or financial risk)is reduced by a certain factor depending on the locationin the quantitative risk matrix (see Table 2). Part of thelikelihood of failure analysis involves assigning levelsof effectiveness for past inspections. The effectivenessof the inspection methods to detect the damage mecha-

nisms is evaluated and characterized based on fiveinspection effectiveness categories: highly effective,usually effective, fairly effective, poorly effective andineffective. Assignment of categories is based on pro-fessional judgment and expert opinion. These cate-gories are applied during the RBI planning. The start-ing point to evaluate different inspection programs is tocalculate the likelihood of failure for the different dam-age states, accounting for the previous inspectionresults and the maintenance history of the equipment.

One of the most important criteria is the capability ofthe inspection methods to detect the characteristics ofthe relevant damage mechanisms. The damage mecha-nisms considered cover:

Table 3. Top Ten High Risk Items

Exchanger

Heater

Vessel

Vessel

2 in. Pipe

4 in. Pipe

2 in. Pipe

2 in. Pipe

16 in. Pipe

16 in. Pipe

Reformer Gas Boiler-Shell

Primary Reformer Tubes

Ammonia Converter

Purge Gas AmmoniaStripper

NH3 Stripper

Syngas Chiller

NH3 SynthesisSecondary Separator

Ammonia Refrigation

CC>2 Absorption

CO2 Absorption

3330

1104

950

204

4

4

4

4

2

1

HTHA

Creep

HTHA

Internal Thinning

Internal LocalizedThinning and GUI

Internal Thinningand GUI

AtmosphericCorrosion

GUI

Internal LocalizedThinning

Internal LocalizedThinning andAtmosphericCorrosion

5

5

4

4

4

4

4

4

4

4

E

E

E

E

E

E

E

E

E

E

AMMONIA TECHNICAL MANUAL 152 2001

• Thinning (external corrosion, corrosion under insu- Study Resultslation (GUI) and internal corrosion).

• Surface-connected cracking. The RBI assessment performed included 144 fixed• Subsurface cracking. pressure vessels (vessels, columns, exchangers, and air• Microfissuring/microvoid formation. coolers) and 180 piping items. The following degrada-• Metallurgical changes. tion models were identified as being active mecha-• Dimensional changes. nisms in the plant: internal thinning, atmospheric cor-• Blistering. rosion, corrosion under insulation, high temperature• Material property changes. hydrogen attack and creep.

Table 4. Inspection Planning RecommendationsEquipment Description: Reformer Gas Exchanger (Carbon Steel Section)Risk Rank: HighLikelihood Driver: HTHA, Highly susceptibleNumber/Effectiveness of Recommended Inspections: 2BDamage Factor without Inspection by 2005: 2000Damage Factor with Inspection by 2005: 800Inspection for HTHA: Maximum interval of 2.5 years to maintain factor of 400. The recommendedinspection technique involves a combination of Backscatter, Velocity Ratio, and Spectrum Analysis, fol-lowed up with field metallography.

Equipment Description: 12 in. Pipe NH3 Synthesis PreheatRisk Rank: HighLikelihood Driver: Atmospheric CorrosionNumber / Effectiveness of Recommended inspections: 1ADamage Factor without Inspection by 2005: 520Damage Factor with Inspection by 2005: 20Inspection for External Thinning only. Visual inspection of >95% of the exposed surface with follow-up by UT, RT, or pit gauge as required.

Equipment Description: 18 in. Pipe in MEA Regenerator SectionRisk Rank: Medium HighLikelihood Driver: GUINumber/Effectiveness of Recommended Inspections: 1BDamage Factor without Inspections by 2005: 400Damage Factors with Inspections by 2005: 110Inspections for GUI: 95-100% external visual inspection prior to removal of insulation. Remove66-94% of total suspect areas and followup of corroded areas with UT, RT or pit gauge.

Equipment Description: 10 in. Pipe in Desulfurization SectionRisk Rank: Medium HighLikelihood Driver: Internal Localized ThinningNumber/Effectiveness of Recommended Inspections: 1ADamage Factor without Inspections by 2005: 290Damage Factors with Inspections by 2005: 20Inspections for Internal Localized Thinning: Inspect for actual thickness with spot UT and update theage and rate of corrosion. Possible UT scanning for localized corrosion.

AMMONIA TECHNICAL MANUAL 153 2001

Likelihood of failure, consequence of failure (loss ofcontainment/leaks) and risk were calculated in terms ofinjury to employees and equipment damage. Current(1999) risk values were calculated and the current riskmatrix is shown in Figure 4. The ammonia plant risklevels are typical in that 10 to 20% of equipment in theammonia and refining industry fall into the high riskcategory. The equipment items that calculated to be inthe high risk category included 15 piping items, twovessels, one exchanger and one heater. Table 3 sum-marizes the damage mechanisms and susceptibility forthe top ten risk items.

The corresponding risk for a five-year look-aheadwas calculated assuming that no inspections would bedone during this time period. The projected future riskswith no inspection are presented in Figure 5. Of the 336pieces of equipment, 10% (32 pieces) of the equipmentitems were projected to rise from an acceptable level ofrisk into the unacceptable level by the year 2005. Aninspection plan was developed that directed the inspec-tion efforts toward items above the unacceptable risklevel. The prioritization enabled the plant to determinethe optimum level of inspection or inspection effec-tiveness required to maintain the risk of equipment atits present level as the equipment ages or to reduce the

AWMO»S»PL«OT-CURR£>iTKai STATUS

its.

Toute

Figure 4. Ammonia plant - current year (1999)RBI status.

risk of equipment when the present level of risk isunacceptable. Table 4 shows some typical inspectionplanning recommendations that resulted from the RBI

Table 5. Current and Future Risk Projections Before and After Inspection

RiskRankings

High

MediumHigh

Medium

Low

Total

Current (1999)

Items

19

185

78

54

336

% of Total

5.65

55.06

23.21

16.07

Year 5 (2005)Before Inspection

Items

24

187

75

50

336

% of Total

7.14

55.65

22.32

14.88

Year 5 (2005)After Inspection

Items

20

181

82

53

336

% of Total

5.95

53.87

24.40

15.77

AMMONIA TECHNICAL MANUAL 154 2001

18

<sa

Figure 5. Ammonia plant - 2005 RBIprojection without inspection.

analysis.Implementation of the inspection planning recom-

mendations developed by DNV together with plantpersonnel will result in a projected risk profile for2005, as shown in Figure 6. After one set of inspec-tions, this ammonia plant had 20 items that remained inthe high risk category. Items which show a high likeli-hood of failure/high consequence even after inspectionmay require repair or replacement. In some instances,when the recommended inspections are done, the addi-tional data gathered may lower the likelihood of fail-ure.

Typically, conservative assumptions are used duringthe analysis which may be tempered by the collectionof better data. This ammonia plant also showed a largenumber of low/likelihood/high consequence items.Since inspection planning only addresses likelihood offailure, reducing the risk associated with these itemswould require additional consequence mitigation or

AMMOK6V ClAKT - 30CS Wfm

Figure 6. Ammonia plant - 2005 RBIprojection with inspection.

detection efforts. Potential detection systems wouldinclude instrumentation specifically designed to detectmaterial losses by changes in the operating conditionsor detectors in areas where toxic materials are handled.These detections systems need to be complementedwith isolation systems that can respond quickly to anyrelease. The RBI analysis also confirmed that pastinspection efforts at the ammonia plant have beeneffective in minimizing the risk to personnel and equip-ment. The benefits of implementing the inspection plancan also be seen in Table 5 which summarizes the cur-rent and future risk projections before and after inspec-tion. The inspection and associated maintenance costsavings were estimated at $237,500 for a Return on theInvestment (ROI) of 5:1. The maintenance cost savingsare based on two factors: a reduced number of instec-tions and the use of nonintrusive inspections when ves-sel entry is not required by operations. The use of non-intrusive inspections substantially reduces the mainte-

AMMONIA TECHNICAL MANUAL 155 2001

the maintenance costs required to prepare a vessel forinspection and may contribute to a shorter shutdown orturnaround. Internal (intrusive) inspections are costlysince they require internal cleaning, blinding, monitor-ing of flammables and toxics inside the vessel, and theuse of a safety watch person at all times. The annual-ized savings were estimated at $47,500, which equatedto a 5% reduction of their maintenance and inspectionbudget.

Conclusions

Risk-based inspection is a subset of a risk-basedmanagement system. It gives management the tools itneeds to make cost/benefit decisions regarding inspec-tion and inspection related maintenance activities.

Risk is a good criterion for prioritizing inspectionefforts because:

• Highest priority items are easily identified as thehighest risk items.

• Risk can be measured in personnel safety and eco-

nomic terms.• Inspection and maintenance activities can be justi-

fied on a cost/benefit basis.API RBI is an effective tool for prioritizing inspec-

tion for the ammonia industry. For likelihood of failureanalysis, damage mechanisms such as localized thinn-ing due to CC>2, HTHA, amine stress corrosion crack-ing, and corrosion under insulation which are active inammonia plants are found in the RBI technical mod-ules. For consequence analysis, the toxic effects ofammonia and fluids typically handled by ammoniaplants are also found of the RBI consequence model.

Literature Cited

API Committee on Refinery Equipment: "BaseResource Document on Risk based Inspection," DNVIndustry Inc., Houston, TX (Jan. 1998).

Madsen, H. O., N. C. Lind, and S. Krenk, Methods ofStructural Safety, Prentice-Hall, Engelwood Cliffs, NJ(1986).

QUESTIONS AND ANSWERS

Ved R. Gupta, Incitée Ltd.: At the IncitéeNewcastle, Australia, ammonia plant, we havedone a similar RBIP. However, before making adecision to bring the scheduled shutdown forward

for physical inspection/repair of a particularequipment, we used acoustic emission monitoring.Did you consider that in your study?

Lily Sweet, Det Norske Veritas (U.S.A.), Inc.:No.

AMMONIA TECHNICAL MANUAL 156 2001