ASSESSMENT OF PRV TURNAROUND INTERVAL Alex Bond, Arcadis Vectra, UK; Adrian Beec, Centrica Energy and Jon Gould, Centrica Energy, UK Te inspection and testing of pressure relief valves (PRVs) is a critical element of te integrity management programme for all process and cemical plant. Routine testing and inspection provides te assurance tat te relief valve will operate correctly wen called upon to do so. However tis level of assurance comes at a price particularly were te process as to be alted in order to provide access to te relief valve, and terefore it is important to set te inspection/test interval at an appropriate interval suc tat te optimum compromise between reliability and lost production is acieved. In general, for te oil and gas industry, te periodicity for inspection and testing of PRVs is governed by te guidance of API 510, wic states tat: Pressure-relieving devices sall be tested and inspected at intervals tat are frequent enoug to verify tat te valves perform reliably in te particular service conditions. It is ten left to te inspection engineer to set te appropriate interval, wit te guidance tat te interval sould not exceed five years for typical process services; and ten years for clean (non-fouling) and non-corrosive services, unless a risk based assessment as been completed. In practice owever, most PRVs are tested and inspected on a bi-annual or tri-annual basis, and terefore tere is a potential for significant cost savings to be made so long as a robust justification can be made, wic demonstrates tat te residual risk remains witin tat considered to be broadly acceptable. Working closely wit Centrica Energy, Arcadis-Vectra as developed a metodology for evaluating te past inspection/test istory for PRVs and, using a risk based approac based on te guidance in API581, to assess te residual risk presented by extending te inspection interval. Te result is a systematic approac to te determination of an appropriate inspection interval for PRVs. INTRODUCTION Almost all businesses associated wit production will ave a number of pressure relief valves (PRV) on site. Tese valves form a key line of defence against te catastropic failure of any pressure system, and as suc are normally regarded as safety critical equipment. Often tey are located externally, in relatively exposed locations, and terefore subject to te ravages of te environment and accidental damage. It is terefore vital to ensure tat tey are in good condition, suc tat in te event of demand tey will operate as intended. Tis is acieved by a programme of testing, inspection and refurbisment. Te periodicity for effective inspection and testing of PRVs is generally governed by fairly generic guidance in standards suc as API 510 [1], wic states tat: Pressure-relieving devices sall be tested and inspected at intervals tat are frequent enoug to verify tat te valves perform reliably in te particular service conditions. It is ten left to te inspection engineer to set te appropriate interval, wit te guidance tat te interval sould not exceed five years for typical process services; and ten years for clean (non-fouling) and non-corrosive services, unless a risk based assessment as been completed. In practice owever, most PRVs are tested and inspected on a simple time-based bi-annual or tri-annual basis. Tis approac owever does not take account of past operating istory, or te criticality of wat te valve is protecting, and can lead to significant unnecessary expenditure, particularly were it is necessary to sut-down te process in order to safely remove te PRV for testing due to over testing, or more importantly, reduced safety due to under testing. Terefore tere is a potential for significant cost savings to be made so long as a robust justification can be made, wic demonstrates tat te residual risk remains witin tat considered to be broadly acceptable. Working closely wit Centrica Energy, Arcadis- Vectra as developed a metodology for evaluating te past inspection/test istory for PRVs and, using a risk based approac based on te guidance in API581, to assess te residual risk presented by extending te inspection interval. Te result is a systematic, robust approac to te determination of an optimum inspection interval for PRVs in wic it is clearly demonstrated tat tere is no significant cange in te risk exposure to personnel and te wider public following te adoption of increased inspection/test intervals. ASSESSMENT METHODOLOGY Te PRV test interval is determined by following a series of steps as sown in te flowcart, Figure 1. Eac stage of te process is described in more detail in te following sections of te report. INITIAL DATA REVIEW Prior to carrying out a risk ranking exercise, te raw inspection/test data needs to be reviewed to identify weter tere are any obvious trends in te data, or particular valves wic ad sown a istory of repeated failure, and terefore sould be treated as special cases in te context of assessing te future inspection regime. Were suc instances are identified, tey sould be te subject of a detailed review and 220
Figure 1. PRV test interval determination 221
consideration as to weter design canges sould be implemented to improve reliability. PRELIMINARY INTERVAL ASSESSMENT Te next stage of te process is to determine an initial assessment of an acceptable test/inspection interval. Te following simple approac as been sown to provide a good basis for setting intervals, and can be sown to give pragmatic results. Te approac as been successfully used at a number of organisations, and provides a simple basis for reduction/increase of inspection interval in te ligt of inspection/test results. INSPECTION GRADE At te eart of te metod is te concept of Inspection Grade, broadly based on te definitions given by te Institute of Petroleum [3] in teir Model Code for Safe Practice. Tese inspection grades, combined wit an assessment of te criticality of te protected equipment, provide a basis for setting te inspection frequencies for individual PRVs, dependent on te findings of previous inspections, or in te case of newly installed equipment, an initial inspection frequency. Te inspection grades are defined as follows: Grade 0 Grade 1 No previous inspection istory available. Allocated to a PRV were tere as been at least one successful turn-around. Inspection as identified some deterioration in PRV condition, wic could ultimately lead to future failure, tus continued frequent inspection required to monitor progress i.e. certain present condition, uncertain future condition. Grade 2 PRVs were successive inspections ave indicated satisfactory reliability, any observed deterioration as been sown to occur at a reasonably predictable rate, consistent wit te inspection period. Grade 3 Allocated to a PRV wen deterioration as been sown to be at a low and predictable rate, service conditions are known, and tere as been a successful inspection following a Grade 2 interval. Following a successful initial toroug inspection (Grade 0 period) te PRV will be allocated a Grade 1, and tence to Grade 2 at te next turn-around, depending on te results of te inspection. A PRV will not be allocated a Grade 3 Inspection Grade until it as demonstrated a successful Grade 2 inspection and as been in service for in excess of 48 monts. In te event tat process conditions are canged, te Inspection Grade sould be re-set to Grade 0 in order to ensure tat consequential canges to degradation rates or fouling potential are not overlooked. Te flowcart in Figure 2 presents te sequence of assessments required to determine weter te inspection grade can be increased at te end of an inspection or not. In essence, if te last two inspections were satisfactory, and te inspection did not igligt a potential problem, ten te inspection grade can be increased. If te PRV failed te current inspection, and te pre-pop pressure was greater tan te set pressure, ten te inspection grade sould be reduced. CONSEQUENCE ASSESSMENT A simplified consequence assessment is used in te PRV ranking, using a qualitative approac. For eac PRV, te Figure 2. Inspection grade allocation flowcart 222
Table 1. Consequence rating Consequence People Business Assets Environment Reputation 1 Catastropic Multiple fatalities or permanent total disabilities Substantial loss of productivity. Extensive damage. 5M 2 Severe Single fatality or permanent Partial loss of Major damage 1M disability productivity. 5M 3 Major Major injury or ealt effects Partial sutdown. Local damage 100k 1M 4 Marginal Minor injury or ealt effects Brief disruption. Minor damage 10k 100k 5 Negligible Sligt injury or ealt effects Negligible Sligt damage disruption., 10k Massive effect Major effect Localised effect Minor effect Sligt effect International impact National impact Considerable impact Minor impact Sligt impact worst, most likely consequence of te failure to operate on demand is assessed on a scale of 1 5, as sown in Table 1. Tis approac is compatible wit te majority of risk based metodologies wic rely on a risk matrix, and many companies already ave appropriate definitions of te consequence categories, calibrated to teir own company and industry expectations. INTERVAL SETTING Based on te above Inspection Grades, and te consequence score for te protected equipment, te following maximum recommended inspection intervals are provided, Table 2. INTERVAL VALIDATION In order to validate te inspection interval determined as above, te risk exposure to personnel on site is determined, and compared against an acceptance criterion. Te metodology tat as been used to carry out te risk exposure calculation is based on te metodology presented in API 581 [2]. Tis approac as been selected as it provides a degree of credibility and level of industry acceptance to te calculation metodology, as well as providing an additional degree of independence to te assessment. Te principles of te assessment are outlined in te flowcart, Figure 3. Consequence score Table 2. PRV test interval setting Maximum recommended interval between inspections (monts) Grade 0 Grade 1 Grade 2 Grade 3 1 6 12 24 48 2 12 24 36 48 3 12 24 48 60 4 24 36 48 60 5 36 48 60 72 LIKELIHOOD ASSESSMENT At te eart of te likeliood assessment metodology is te use of a two parameter Weibull distribution, wic gives te cumulative probability of failure to open on demand F(t) as: " F(t) ¼ 1 exp t # b were b is te sape function, and is te caracteristic life. Te values used for and b in te assessment can be derived from site specific information, or be taken from Table 7.5 of API 581 for PRVs of te appropriate type and service fluid type. Te advantage of using te API tabulated values is tat te information is broadly accepted witin industry wit no furter justification, and is conservative. Te caracteristic life parameter is ten adjusted based on te prior inspection istory results using a Bayesian approac, were te caracteristic life parameter is revised using equation 2. (1) t upd ¼ (2) 1 ln 1 p prd b f,wgt Te full API assessment also considers te likeliood of PRV leakage; owever for te purposes of tis simplified assessment, leakage as not been considered, as te consequences are minor compared to tose due to failure of te valve to open on demand. Te inspections carried out by Centrica Energy ave included pre-pop-testing, and terefore are considered to be igly effective (in te context of te API 581 definition), wit a 90% probability of identifying a faulty PRV. Te weigted probability of failure on demand is terefore calculated as: p prd f,wgt ¼ p prd prd f,prior 0:2 p f,prior t þ 0:2p prd f,cond t (3) 223
SYMPOSIUM SERIES NO. 156 Hazards XXII # 2011 ICemE Figure 3. API metodology flowcart 581 as been used directly. For eac overpressure demand case, an associated event frequency is provided, based on industry experience (Table 7.2 of API 581). Te likeliood of te protected equipment being damaged as a consequence off any particular overpressure event is ten determined. Tis is based on a consideration of te current equipment condition, te ratio of likely over pressure to design pressure and te generic failure frequency for te equipment type. For reasons of simplicity, for te current evaluation, all protected equipment as been given a prior damage rating of Minor ; tat is inspection is routinely carried out and as confirmed one or more damage mecanisms is active, owever damage is witin expected limits, and unlikely to result in premature equipment for a successful inspection/test, and prd p prd f,wgt ¼ p f,cond (4) for a failed inspection/test. Tis evaluation is repeated for eac PRV, for eac of te prior inspections, and a revised caracteristic life determined for eac valve. At te end of tis process, te probability of failure on demand for eac valve at te end of te preliminary test interval specified at te start of te process is determined. Te next stage of te API assessment metodology is to determine te likely demand rate for eac valve. Again, for reasons of ensuring credibility, te guidance of API 224
failure. In te context of te Sout Morecambe plant, tis is considered to be a conservative basis for te majority of equipment, and ence a reasonable basis on wic to develop te metodology for PRV inspection planning. Te adjusted failure frequency due to overpressure is ten calculated using equation 5, as follows: P f ¼ (P gen DF) þ 1 P gen P o 3 MAWP 1 were P gen is te generic failure frequency for te equipment, P o is te anticipated over-pressure due to failure of te PRV, MAWP is te maximum allowable working pressure (in te context of te current assessment tis is taken as te design pressure) and DF is te damage factor wic in tis assessment as been taken as 200 trougout, based on an assumed damage grading of Minor. Te generic failure frequency is taken from Table 4.1 of API 581 for te rupture case. Te overall likeliood of an accident occurring due to failure of a PRV to operate on demand is ten calculated as: Likeliood ¼ Probability of failure to operate on demand Demand Rate Probability of failure due to over-pressure SIL Adjustment Factor Population Adjustment: Te SIL Adjustment factor is intended to take account of iger SIL rated controls resulting in a lower demand rate on te PRVs. Te adjustment factors used are sown in Table 3. Te population adjustment modifies te calculated likeliood to take account of te normal level of occupancy in te vicinity of te PRV; in line wit oter risk based metodologies, te cut-off point for considering a location to be occupied is taken as over 10% of te time. RISK EXPOSURE ASSESSMENT Te risk exposure level is considered on te context of te HSE R2P2 guidelines [4] for ALARP as sown in Figure 4. For te most critical incidents, te likeliood of an incident occurring must remain below 10 26 to drop into te broadly acceptable region, wilst a likeliood of over 10 23 would be considered intolerable (see Table 4). As te anticipated worst-case consequence reduces, so te Table 3. SIL adjustment factor SIL level Factor NR 1.0 SIL 1 0.1 SIL 2 0.01 SIL 3 0.001 SIL 4 0.0001 (5) Consequence rating Table 4. Risk exposure tolerability Broadly acceptable ALARP Intolerable 1,1.00E-06 1.00E-6, L, 0.001.1.00E-03 2,1.00E-05 1.00E-5, L, 0.01.1.00E-02 3,1.00E-04 1.00E-4, L, 0.1.1.00E-01 4,1.00E-03 1.00E-3, L, 1.1 5,1.00E-02 1.00E-2, L, 10.10 tolerable likeliood is increased. For convenience, a logaritmic scale as been used, wit te tolerable frequency increasing by an order of magnitude for eac drop in consequence category. Te following break-points ave been used for te PRV assessment. Were te risk exposure due to te preliminary test interval determined in Step 1 remains witin te Broadly Acceptable region, te interval can be considered to be acceptable. In te event tat te interval results in te risk exposure being in te ALARP region, ten a cost-benefit assessment sould be carried out. In most instances tis would suggest tat a reduced inspection interval would be more appropriate. Any resulting in te risk exposure being in te intolerable region sould as a matter of course be tested/inspected on a more frequent basis. It is likely tat te overall strategy for pressure control in suc examples sould also be reviewed, as te risk reduction expected by te PRV is greater tan would normally be acceptable. CONCLUSIONS Te metodology described as been applied to bot onsore and off-sore pressure relief valves at Centrica Energy s Morecambe facilities. For te majority of PRVs, te metodology as sown tat te inspection/test interval can be increased by a year witout moving te residual risk exposure from te Broadly Acceptable region of te tolerability grap; te result is significant cost savings and improved production flexibility. Te next stage of te investigation is to monitor te results from te next inspection test campaign in tree to four years time, to see if tere as been an increase in te number of PRVs failing te pre-test. Assuming tat te test failure rate does not increase from previous campaigns, ten te approac will ave been vindicated. Initial conservative estimates, based on labour costs only, indicate cost savings to Centrica of around 100,000 savings over te remaining life of te facility. Clearly tere are furter potential savings associated wit reduced plant down-time and access (scaffolding), wic ave not been included in te above estimate. For oter facilities owever, te actual cost of saving acieved will be dependent on a number of factors; weter te PRVs ave been designed wit parallel valves to allow on-line removal of valves for testing; accessibility, current inspection regime, cost of lost or deferred production. 225
Figure 4. Tolerability of risk Altoug te metodology as been developed specifically for PRVs, it could be modified to be used for oter safety equipment were periodic inspection and re-calibration is required, for example flue gas analysis detectors etc. GLOSSARY ALARP API MAWP PRV SIL b As Low As Reasonably Practicable American Petroleum Institute Maximum Allowable Working Pressure Pressure Relief Valve Safety Integrity Level Sape function Caracteristic life function DF F(t) Damage factor Probability of failure to open on demand REFERENCES 1. American Petroleum Institute, 2006, Pressure Vessel Inspection Code: In-Service Inspection, Rating, Repair, and Alteration, API 510, 9t Edition. 2. American Petroleum Institute, 2008, Risk-Based Inspection Tecnology, API Recommended Practice 581, 2nd Edition. 3. Energy Institution, 1993, Model Code of Safe Practice in te Petroleum Industry Pressure Vessel Examination, IP12. 4. Healt and Safety Executive, 2001, Reducing Risk, Protecting People, ISBN 0 7176 21 51 0. 226