A method for developing and structuring risk activity indicators for major accidents

Transcription

1 A method for developng and tructurng rk actvty ndcator for major accdent Jan Erk Vnnem Preventor/Stavanger Unverty College, Bryne, Norway Gunnar Vere Statol, Stavanger, Norway Bjørnar Hede Knuden Safetec Nordc, Stavanger, Norway Terje Aven Stavanger Unverty College, Stavanger, Norway Abtract An actvty ndcator for work on offhore petroleum nta l- laton developed, n order to reflect the rk of hydrocarbon leak durng planned actvte n a medum long tme cale. The ndcator may be ued proactvely durng pla n- nng, and are rk baed, n the ene that the rk level reflect the lkelhood of lo of peronnel, accordng to the everty of the leak, and expected performance of barrer. The approach brefly compared to a more mplfed cheme. Further work outlned, ncludng extended focu on concurrency of actvte, and tetng on a plot ntallaton. 1 Introducton The Rk level report for the Norwegan Contnental Shelf [1] ha documented that only about 30% of the hydrocarbon leak are due to normal operaton, mplyng that the majorty of the leak are due to ome knd of manual actvte, nterventon and mantenance work. Hydrocarbon leak are uually the man ource of rk for major accdent for the offhore petroleum producng ntallaton. It therefore mportant to addre rk aocated wth manual actvte. Th the topc of a reearch actvty fnanced by Statol and the Norwegan Reearch Councl, called Actvty ndcator. Statol ha been developng four dfferent ndcator for major hazard rk [2]: Bac barrer ndcator Barrer performance ndcator Incdent ndcator Actvty ndcator

2 Actvty ndcator are partcularly mportant, a they are proactve ndcator, a the other ndcator are reactve. The man beneft of the proactve ndcator that they are forward lookng, and may be ued for optmzaton and plannng purpoe. Clearly t ueful to combne reactve and proactve ndcator. The paper preent the current work n the reearch project n order to develop actvty ndcator for Statol, at preent lmted to actvte that may caue hydrocarbon leak, baed on work n [3], [4], [5] and [6]. The ba for the development to a large extent npred by work n the nuclear ector, ee for ntance [7], [8] and [9]. 2 Decrpton of Approach The approach decrbed n th ecton a bref ummary of the approach. The full decrpton avalable n [10]. An actvty ndcator may be tated a AI = n w (1) - AI a rk baed actvty ndcator for an actvty, - n the number of planned actvte for actvty n a future perod, and - w a rk weght factor for actvty,.e. the expected rk per actvty The uer of the actvty ndcator expree the actvty level n accordng to the actvty plan. We expre the weght factor w a ( w = P L ) E( X L ) (2) - the ze of the hydrocarbon leak - L the occurrence of a hydrocarbon leak - a rk actvty, and - X the number of fatalte. Incdent and expoure data for one Statol operated feld durng a fve-year perod were examned. The dtrbuton of leak ze are quantfed a P( L ) = P( L ) ϕ φ (3) - - ϕ expree the oberved proporton of gnfcant hydrocarbon leak related to actvty on the Statol operated feld. φ expree the oberved proporton of gnfcant actvty related hydrocarbon leak that fall n the categore on the Norwegan Contnental Shelf n the perod A entvty analy wa carred out to decde whch of the analyed actvte that have the greatet potental for rk reducton. The weght factor w erve a an expreon of entvty. Baed on the entvty analy t wa concluded that the actvty ndcator hould be baed on Work on preurzed equpment, Start-up of proce equpment and Change n proce condton.

3 2.1 Preentaton of the actvty ndcator A bubble dagram well uted to preent nformaton n three dmenon. In the fgure below each actvty repreented by a dfferent colored bubble. The y-ax repreent a poble actvty level aocated wth the actvte, and the ze of the bubble repreent the actvty ndcator value aocated wth the actvty level of each actvty. Actvty level Work on preurzed equpment Change n proce condton Start-up of proce equpment Fgure 1 Actvty ndcator aocated wth a typcal actvty level durng a year on a large offhore petroleum ntallaton Th fgure clearly how how the actvte contrbute to the rk ndcator pcture. The actvty planner wll wth uch a tool have an opportunty to evaluate the plan. He/he can change the number of work order related to each actvty, and get a quck overvew of how th wll change the ze of the actvty ndcator bubble. Th could be a valuable tool to reduce rk at the ntallaton. The actvty planner may alo be ntereted n gettng an overvew of the actvty ndcator durng a hgh actvty perod. To facltate th, accumulated ndcator value for each week could be preented. Another mportant feature of the actvty ndcator that t could be ued to chooe how a project hould be mplemented, by comparng the actvty ndcator core of dfferent alternatve. 2.2 Ue of the Actvty Indcator Statol could be ntereted n makng the actvty ndcator a tandard procedure for key peronnel at the ntallaton. Many actvte are uually planned on a weekly ba, and then more n detal on a daly ba. When the actvte are planned, thee plan wll be regtered electroncally n the control room, o th a natural tme to ue the actvty ndcator. The actvty ndcator could be made avalable for the management. The manager may then tet ther actvty plan and combnaton, before plan are fnalzed. Th may help avodng unwanted hgh-rk perod, and could alo erve a a tool to boot nvolvement n and undertandng of mportant afety ue. A preadheet ha been created to facltate the ue of the actvty ndcator.

4 2.3 Data Ba Incdent and expoure data for ten actvte and condton were gathered, and were ued to expre the leak probablte that the actvty ndcator are baed on. The choen approach take nto account that the dtrbuton of leak related to Change n proce condton qute dfferent from the two other actvte. 3 Decon Support and Challenge A proactve ue of the ndcator make t poble to nclude rk mpact n actvty plannng, n partcular when decdng to mplement rk reducng meaure. Indcator value for dfferent et of planned actvte can be compared. They can be plotted a a functon of actvty type, tme or dfferent project opton, gvng the decon maker the poblty to put afety related focu rk hghet (ee example n Fgure 1) The ndcator model preented n [3] focue prmarly on rk related to ndependent actvte. Set of actvte can be evaluated and compared wth regard to hydrocarbon leak rk, wth a ba n the local rk analy. However, the actvty Hot work, and the condton Backlog of planned mantenance, repreent rk only n concurrency wth another event; a leak. Hence, the relevant weght factor for Hot work and Backlog of planned mantenance refer to the combnaton of uch actvte and hydrocarbon leak. A more complcated apect of multaneoune preently not ncluded n the actvty ndcator model: How wll a number of parallel actvte nfluence rk? Intutvely, Work on preurzed equpment multaneouly wth Contructon work condered to have a hgher rk than the ame actvte performed n equence. Expreng the rk related to hgh actvty level (a hgh number of parallel actvte) dffcult, and there are lttle data avalable for etablhng weght factor. The ue of parallel actvte hould be further explored, a t would gve decon upport to rk evaluaton performed daly by operaton peronnel: How to coordnate actvte n a way that mnmze rk? 4 Alternatve Approach In the ol and ga ndutry alo other approache for ung actvty baed ndcator have been condered. An example the approach by Nork Hydro, n whch actvty ndcator are defned and ued a a part of a larger group of ndcator related to major hazard. From an ntal et of actvte that nfluence rk: correctve mantenance, hot work, crane lft, backlog of preventve mantenance, outtandng correctve mantenance, outtandng meaure, alarm, hutdown, electrc power falure and nhbt (nhbt of e.g., ga detector), three were cho-

5 en to be the ba of Nork Hydro actvty ndcator: Backlog of preventve mantenance, Outtandng correctve mantenance and Inhbt. The choce wa made baed on the actvte gnfcance for afety and the acce of data. An overall key performance ndcator related to actvte, KPI A, ummare the nformaton from the varou ndcator. Th overall ndcator defned by 3 k1 N PM + k2 NCM + k3 N BP k N KPI A = = (4) n n = 1 3 k = 1 - k a rk weght factor for each actvty, = 1 - N PM the number of work order related to preventve mantenance - N CM the number of work order related to correctve mantenance - N BP the number of nhbt and by-pae n the lat three month, and - n the number of tag. The goal of the ndcator to mprove the afety management by montorng the techncal condton. The ndcator are ued by all ector of Nork Hydro, and are baed on extng reportng ytem,.e. by countng equpment and revewng databae. The ndcator are ued by the manager and afety npector, and gve a focu on afety, dentfy need for change and mprovement. They are alo ued for budget prortzaton. It ha recently been decded to remove N CM and N PM from the orgnal actvty ndcator a expreed by KPI A above. Th mple that only nhbt reman a the ba for the ndcator. 5 Concluon and Recommendaton Actvty ndcator may be ued to optmze actvty plan and avod hgh-rk perod. The approach preented n th paper relatvely detaled and requre qute extenve nput data. Th neverthele the recommended approach due to the opportunty to nfluence plannng n a proactve manner. A plot project beng planned n order to tet the approach on a pecfc ntallaton. Th wll provde more nght nto whether and how the actvty ndcator hould be mplemented on a permanent ba, ncludng: For whch plannng horzon would the actvty ndcator be mot ueful? (Long-term plannng; 1-3 month; 1-7 day?) How can concurrency between actvte bet be ncluded? Should Hot work or other actvte be ncluded? How ha the plot ntallaton benefted from the ue of actvty ndcator? It alo the ntenton to tet the mple, reactve ndcator, baed on the number of nhbt retropectvely, n order to learn what th approach can offer a a upplement to the proactve ndcator.

6 Acknowledgement The work of Cathrne Torjuen n her mater the wa an mportant ba for the work. The ubject ha alo been dcued wth Unn N. Samdal, Nork Hydro Porgrunn, who ha provded an nteretng alternatve perpectve. Both have provded valuable nput to the work, and are gratefully acknowledged. The project fnanced by the Reearch Councl of Norway and Statol, whoe contrbuton are gratefully acknowledged. Reference 1 NPD. Trend n Rk Level on the Norwegan Contnental Shelf, Summary Report, Phae , Report OD-03-08, Thomaen, O., Sørum, M. Mappng and montorng the afety level, SPE 73923, preented at SPE Internatonal Conference on Health, Safety and Envronment n Ol and Ga Exploraton and Producton, Kuala Lumpur, March, Hede Knuden, B. Proactve rk ndcator: Actvty ndcator for major hazard rk on offhore ntallaton (n Norwegan), Mater The, Stavanger Unverty College, Torjuen, C. Actvty ndcator for ga leak whch data are avalable and how can they be ued (n Norwegan) Mater The, Stavanger Unverty College, Aven, T. Foundaton of Rk Analy: A Knowledge and Decon-Orented Perpectve. Wley, New York, Vnnem, J.E. Rk Montorng for Major Hazard, SPE61283, SPE Internatonal Conference on Health, Safety and the Envronment n Ol and Ga Exploraton and Producton n Stavanger, Norway June IAEA. Lvng Probabltc Safety Aement. IAEA-TECDOC-1106, IAEA. Indcator to montor NPP operatonal afety performance. IAEA-J4-CT-2883, Shepherd, C.H. Rk Montor, A report on the tate of the art n ther development and ue, Produced on behalf of OECD WG rk and IAEA, Compled by Dr C H Shepherd, HM prncpal npector, Nuclear ntallaton npectorate, UK 10 Vnnem, J.E,, Aven, T., Vere, G., Hede Knuden, B. Actvty ndcator for major hazard rk. To be ued, 2004