The Importance of Real-Fluid Behaviour in Predicting Release Rates Resulting from High Pressure Sour-Gas Pipeline Ruptures

The Canadian Journal of Chemical Engineering (Impact Factor: 1.23). 02/1989; 67(1):3 - 9. DOI: 10.1002/cjce.5450670102


The real-fluid nonisentropic decompression model presented by Picard and Bishnoi (1988) has been used to evaluate the current practice of assuming perfect-gas behavior when predicting release rates from high-pressure sour-gas pipelines. It is shown that use of perfect-gas theory can result in the transient release rate being underestimated by 30 to 45 percent, and the total amount of fluid released, underestimated by 50 percent. Furthermore, perfect-gas theory cannot consider condensation effects; more than 35 percent of the H2 S contained in a pipeline may be emitted as a liquid.
Use of the popular “double exponential” model, as presented by Wilson (1979), will result in similar errors. However, if the model is modified slightly so that the initial mass of fluid in the pipeline is determined based on real-fluid theory, then a conservative estimate of the transient release rate may be obtained (i.e., the most rapid release of the fluid is predicted).
On a utilisé le modèle de décompression non-isentropique de fluide réel présenté par Picard et Bishnoi (1988) pour évaluer la pratique courante qui consiste à supposer un comportement de gaz parfait dans la prédiction des vitesses de relǎchement des pipelines de gaz corrosifs à haute pression. On montre que l'utilisation de la théorie des gaz parfaits peut entraǐner la sous-estimation de 30 à 45% de la vitesse de relǎchement transitoire et de 50% de la quantité totale de fluide relǎché. En outre, la théorie des gaz parfaits ne permet pas de prendre en compte les effets de condensation; plus de 35% du H2 S contenu dans la pipeline peut ětre émis sous forme de liquide.
L'utilisation du modèle répandu “exponentiel double”, tel que présenté par Wilson (1979), peut entraǐner des erreurs similaires. Cependant, en modifiant légèrement le modèle de telle sorte que la masse initiale du fluide dans le pipeline soit déterminée à partie de la théorie du fluide réel, on peut alors obtenir une estimation conservatrice de la vitesse de relǎchement transitoire (soit la prédiction de la plus grande vitesse de relǎchement du fluide).

7 Reads
  • Source
  • [Show abstract] [Hide abstract]
    ABSTRACT: An efficient numerical simulation (CNGS-MOC), based on the method of characteristics for simulating full bore rupture of long pipelines containing two-phase hydrocarbons, was developed. The use of curved characteristics, in conjunction with a compound nested grid system, as well as a fast mathematical algorithm, lead to a significant reduction of CPU time, while improving accuracy. The model is validated extensively against field data including those obtained during the Piper Alpha tragedy, as well as the Isle of Grain depressurization tests. Its predictions are compared with those based on other mathematical models including PLAC, META-HEM, MSM-CS, as well as BLOWDOWN. Both CNGS-MOC and META-HEM produce reasonably accurate predictions with the remaining models assessed performing relatively poorly.
    AIChE Journal 06/1999; 45(6):1191 - 1201. DOI:10.1002/aic.690450605 · 2.75 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The characteristics method was employed to elucidate the effects of fluid-phase transition on the dynamic behavior of emergency shutdown valves following full bore rupture (FBR) of long pipelines containing high-pressure hydrocarbons. The responses of both check valves and ball valves were simulated. The application of the model to the FBR of the Piper-Alpha MCP-01 gas riser reveals that condensation of the gaseous inventory results in a delay in valve activation times, as well as an increase in the amount of inventory released prior to pipeline isolation. However, the magnitude of the pressure surge brought about from having to bring the high-pressure escaping fluid to rest following emergency isolation was significantly reduced.
    AIChE Journal 05/2000; 46(5):997 - 1006. DOI:10.1002/aic.690460512 · 2.75 Impact Factor
Show more