12(3) 10.fm

Transcription

1 KIGAS Vol. 12, No. 3, September, 2008 (Journal of the Korean Institute of Gas) ü LP ƒ š Database w š Á Á½z w ywœw ( , (1 ), (2 ), k) Constructing a Database Structure for the Domestic LP Gas and Natural Gas Accidents and its Analysis Jae-Sun Ko Sun-Young Park Hyo Kim Department of Chemical Engineering, University of Seoul, Korea (Received 10. June. 2008, Revised(1st) 12. August. 2008, Revised(2nd) 5. September. 2008, Accepted 5. September. 2008) 1991 l ¾ 16 3,593 ƒ š[ ƒ (NG) y ƒ (LPG)] w Database w, š xk w. r š xk, s, š y w. š e y, s y, š ƒ šƒ w ùkû. w y, s, ƒƒ Poisson w wz 5 ü y, s w ƒ ƒ y dw. wz w üƒ š Database Á w ü ƒ š d w œw z ƒ» w». Abstract We have garnered 3,593 data of gas [Natural Gas (NG) and Liquefied Petroleum Gas (LPG)] accidents reported for 16 years from 1991, and then analyzed the accidents according to their types and causes based on the classified database. According to the results the gas leak has been the most common accident followed by the explosion and then fire accidents. The most frequent accident-occurring locations for fire, explosion and leak are recognized around the valve, hose and pipeline, respectively. In addition, we have applied the Poisson analysis to predict the most-likely probabilities of fire, explosion and release in the upcoming 5 years. From this research we have assured the successive database updating will highly improve the anticipating-probability accuracy and thus it will play a key role as a significant safety-securing guideline against the gas disasters. Key words : database structure, gas accidents, poisson probability analysis, PDF (probability density function), CDF (cumulative density function) I. ƒ (NG LPG) rwš Áòw 1980 z wš p y yƒ y w ƒ. ƒ w y, s, š ƒ ù jš ƒ w y :119kjs@hanmail.net, s, šƒ wš ƒ œ w w kœ ƒ š ƒ, ƒ y s, š x yw ³ w wz ƒ w w ƒ w y ƒ (Liquefied Natural Gas : LNG) y ƒ (Liquefied Petroleum Gas : LPG) r 56

2 ü LP ƒ š Database w wš Áòw» þ, ù w f. ù ƒ ƒ š ƒ s w ù, w. ƒ ³ ³ w y s x» w x ¾ ƒ s š y š w yw l w š wz w š ù dv w ƒ w w ƒ z w wš üs w x š w œ ƒ e y s vw w š w sƒ» w. p, ƒ w y, s, š l y w wz w š dw v w ƒ w y s š y y w Fig. 1 ww» w l 2006 ¾ ƒ š [1,2] w w yw l w. l y w Poisson s[3] w p p w s³ w p z w y w, ƒ š x š dw v w. II. ƒ š l 2.1. xk ƒ š Fig. 2 ƒ š(ng, LPG, LPG + Air) xk w w ƒ š ƒ w š, w w š. y, s,,», y w w. Fig. 3 ƒ š(ng, LPG, LPG + Air) w w ƒ š 362, ù š 238 ƒ m šƒ ƒ û. Fig. 2. The number of accidents according to the accidenttypes. Fig. 1. Research scheme. Fig. 3. The number of accidents according to the accidentcauses. 57 w ƒ wz 12«3y

3 š Á Á½z 2.2. NG LPG(LPG, LPG + Air) ƒ š LPG(LPG, LPG+Air) ƒ š Fig. 4 ùkù s,, y w šƒ w ùk û, p s w š 495 w š, ƒ š e y,»,, ƒ šƒ w. y w ƒ š 207 w ƒ š e y,», ƒ šƒ w. w ƒ š»,, y, ƒ šƒ w. Fig. 5 NG ƒ š w 547, y, s w p 471, š e,, y w š, y, y,, s, y, ƒ šƒ w œ ƒ š Fig. 6 œ ƒ š w w œ w ƒ š 11, œ 102, w œ 48, œ 68, œ 21, ew 17 ùkû. p 1996» ƒw ƒ š ƒ 2002 œ šƒ š. w Fig. 7 œ š x k œ y xk, œ, œ, w œ, w œ, ew w ƒ šƒ w. Fig. 4. The number of LPG (LPG, LPG + Air) accidents according to the accident-types. Fig. 6. The number of accidents according to the third party excavation-types. Fig. 5. The number of NG accidents according to the accident-types. Fig. 7. The number of gas accidents according to various third party excavation-types. KIGAS Vol. 12, No. 3, September,

4 III. POISSON s w y w ü LP ƒ š Database w 3.1. Poisson s[3] Poisson s p p w s ³ w p z w y ùkü s. p w p w x s Poisson s š w. Poisson s w s y w s³ λ p z (X) w y œ w w. P X = x λ ( ) = λx e λ 3.2. Poisson s y Poisson s w w w ùkü λ( / ) w (1) xw txw. λt r e ( λt) P = r! X!» P (Intensity) λ t r w y w, P (3) λ, t, š r w ƒ. (1) (2) P = P(λ, t, r) (3) wr ƒ λ š š» t w, w Poisson s (4) r w ùký. P = P(r) (4) (4) w R y P y w» w λt 10 l 65¾ 5 ƒ j Fig. 8. v, λ tƒ P w R w., w R R M wš P P M š w, R M P M λt w ƒ. P r r = R M = 0 R M P M λt(ƒ ) w t ù kü Fig. 8. (5) 59 Fig. 8. Locus of Pm-Rm in semi-log scale as the intensities grows. IV. ƒ š DB w NG, LPG š ƒ d 16 ( ) w ƒ œ š 3,593 ƒ [ ƒ (NG) y ƒ (LPG)] š w [1,2]. ƒ ƒ 3 ƒ (NG, LPG + Air), y ƒ (LPG), š ƒ wš ù w ƒ š(ng, LPG, LPG + Air), LPG(LPG, LPG + Air) w š, NG w š 3ƒ w w š, š xk y, s, 3ƒ w š e, y,,» 4ƒ w š ƒ š, e, ³e, œ,, œ 6ƒ w w Poisson s[3] w PDF(Probability Density Frequency)[4] w CDF(Cumulative Density Frequency)[4] MTBF(Mean Time Between Failures)[5,6] w. p Semi-Log t w ƒ¾ xk ƒ l ù w s q w y w w œwš w. ƒ š (y ) LPG LPG+Air(s ), NG( ) ƒƒ e w ƒ w ƒ š y w ƒ š(ng, LPG, LPG + Air) y w ƒ š ƒ š(ng, LPG, LPG + Air) y w ƒ wz 12«3y

5 š Á Á½z Fig. 9. The number of gas accidents according to Total gas (NG, LPG, LPG + Air)-Fire-Pipeline-Sub causes. w 189 w š, y - - œ w š 23, y - - w š 17, y -»- w š 64, y -y - w š 20 w š, Fig. 9 ƒ -y - - w š r ƒ š y - -, y - -, y - - e, y - - ³e ƒ š(ng, LPG, LPG + Air) y - Fig. 10. PDF and CDF for the Fire - Pipeline - Joint loosening event. Fig. 11. Locus of Pm-Rm in semi-log scales by using Poisson distribution of detailed factors of total gas (NG, LPG, LPG + Air) accidents (Pm = Cumulative, Rm = No. of occurrences). - w Poisson s ƒ š(ng, LPG, LPG + Air) y - - w MTBF 1.14(yr/event), Poisson s Fig. 10 wz 5 ü w PDF 5z w y 0.17, CDF 5z w y ƒ š(ng, LPG, LPG + Air) y w Possion s w Fig. 11 ƒ š(ng, LPG, LPG + Air) y w PDF Semi-Log t PDF y - - e (A), y - - e (B), y -y - e (C), y - - (D), y -»- (E), y -y - (F), y - - œ (G), y - - (H) ùkû LPG š(lpg, LPG + Air) s w ƒ š LPG š(lpg, LPG+Air) s w š 495 w š, s - - e w š 31, s - - e w š 59, s -»- w š 48, s -y - e w š 60 w š, Fig. 12 LPG(LPG, LPG+Air)-s - - w ƒ š r ƒ š s - - e, s - -, s - -œ, s - - KIGAS Vol. 12, No. 3, September,

6 ü LP ƒ š Database w Fig. 12. The number of gas accidents according to LPG (LPG, LPG + Air)-Explosion-Valve-Sub causes. Fig. 14. Locus of Pm-Rm in semi-log scales by using Poisson distribution of detailed factors of LP gas (LPG, LPG + Air) accidents (Pm = Cumulative, Rm = No. of occurrences). Fig. 15. The number of gas accidents according to NG- Release-Hose-Sub causes. Fig. 13. PDF and CDF for the Explosion - Valve - Bad finishing event LPG š(lpg, LPG + Air) s - - e w Poisson s LPG š(lpg, LPG + Air) s - - e w MTBF 3.2(yr/event), Poisson s Fig. 13 ùkù wz 5 ü w PDF 10z w y 0.01, CDF 10z w y LPG š(lpg, LPG + Air) s w Poisson s w Fig. 14 LPG š(lpg, LPG + Air) s 61 w PDF Semi-Log Scale PDF s -»-œ (A), s - - œ (B), s - - (C), y - - e (D), s -y - (E), s -y - (F), s - -œ (G), s -y - (H), s -»- (I) ù kû NG š w ƒ š NG š w 228 w š, - - e w š 11, - - w š 9, -y - w š 12 w, Fig. 15 NG- -y - w š ƒ š -y -, -y - w ƒ wz 12«3y

7 š Á Á½z Fig. 17. Locus of Pm-Rm in semi-log scales by using Poisson distribution of detailed factors of NG accidents (Pm = Cumulative, Rm = No. of occurrences). Fig. 16. PDF and CDF for the Release - Hose - Joint loosening event NG š -y - w Poisson s NG ƒ š -y - w MTBF 1.33(yr/event), Poisson s Fig. 16 ùkù wz 5 ü w PDF 5z w y 0.15, CDF 5z w y NG š w Poisson s w Fig. 17 NG š w PDF Semi-Log Scale PDF s - - (A), -y - (B), - - (C), - - Fig. 18. Poisson distribution of total gas accidents. KIGAS Vol. 12, No. 3, September,

8 ü LP ƒ š Database w ³e (D), - - (E), - - œ (F), - - (G) œ (Accidents by Third Party Excavation) sww ƒ š(ng, LPG, LPG + Air) Poisson s w Fig. 18 œ sww ƒ š(ng, LPG, LPG+Air) w Poisson s[3] w v. wz 5 ü(t = 5) š 1) NGs - -, 2) NG- - -, 3) NG- - -, 4) LPG (LPG, LPG + Air)- -»-, 5) NG- - - e, 6) NG- - - ³e, 7) NG- -y -, 8) LPG(LPG, LPG+Air)- -y -, 9) NG-y - - œ, 10) NG- œ - -, 11) NG- - - œ, 12) NG- - -, 13) LPG(LPG, LPG + Air)- -y -, 14) LPG(LPG, LPG + Air)- -»-, 15) NG-w - -, 16) LPG(LPG, LPG + Air)- - -, 17) LPG(LPG, LPG + Air)- œ - -, 18) NG- - -, šƒ w. V. 16 ƒ š [1,2](NG, LPG, LPG + Air) w š xk, s, y w š, š e -y -»-, s w y -»- -, y -y -»-. p w œ - -, œ z š ƒ š, w œ - - ew z œ š» ƒ» y w š z. ù zy š w x w w», m w ƒ š š w. w ƒ w y Ás š w l y Á g, w t g y. w 2007 w (6 ) q w, (½z ). š x [1] ƒ œ, ƒ š, ƒ œ š, ( ) [2] ƒ œ, š ƒ m, ƒ œ, (2001) [3] Robert L. Winkler and William L. Hays, Statistics Second Edition Probability, Inference, and Decision, Holt, Rinehart and Winston, New York, (1970) [4] Roger D. Leitch, Reliability Analysis for Engineers, Oxford University Press, (1995) [5] Calabro, S.P. Reliability Principles and Practices, McGraw-Hill, (1982) [6] Thomas, H.M. Pipe and Vessel Failure Probabilistic Reliability Engineering, Applied Science Publishers, Ltd., England, (1981) 63 w ƒ wz 12«3y