Study of optimal layout based on integrated probabilistic framework (IPF): Case of a crude oil tank farm

https://doi.org/10.1016/j.jlp.2017.04.025Get rights and content

Highlights

  • An integrated probabilistic framework is applied in the optimal layout of a crude oil tank farm.

  • Probability of an initiating event due to self-heating of iron sulfides is determined.

  • Equation deduced to calculate thermal radiation flux is given.

  • Lumped damage risk method is developed and confirmed.

  • Application of lumped damage risk in the optimal layout of a crude oil tank farm certificates its highly effectiveness.

Abstract

This paper gives an integrated probabilistic framework (IPF) that deals with the optimal layout of facilities in an industrial plant. The specific case of a crude oil tank farm is detailed in this present paper, which includes the tank fire as well as the corresponding optimal layout based on inherent safety and evacuation.

The tank fire can be caused by the oxidative self-heating of pyrophoric iron sulfides which extensively exist on the inner wall of the crude oil tank, especially in the respiratory/safety valves. Oxidative self-heating, or spontaneous combustion of iron sulfides is a process of oxidation and generally influenced by five external factors including water content, mass per unit area of iron sulfides, operating temperature of tank, flow rate and concentration of oxygen facing the iron sulfides. According to the previous literature about self-heating process of iron sulfides, the maximum temperature (Tmax) of solid phase is a vital indicator representing the pyrophoric feature of iron sulfides in specific circumstances. And the maximum temperature (Tmax) can be predicted by the model developed from support vector machine (SVM) technique. While the predicted maximum temperature (Tmax) is compared with a defined threshold value, it can be revealed whether the oxidative self-heating of iron sulfides will lead to explosion and then cause tank fire. On this grounds, the probability of tank fire due to the oxidative self-heating of iron sulfides can be obtained by Monte Carlo simulations.

For tank fire, the major physical damage to the surrounding tanks and workers is thermal radiation rather than overpressure or missile projection. Considering the worst case scenario, that is the vapor-liquid interface in the tank covered with fire, then the thermal radiation flux passing through a receiver at a specified distance away from the tank can be derived. In reverse, the critical horizontal distance between tank and receiver can be obtained if the critical thermal radiation flux through a receiver is given. Assuming that the minimum and maximum risks of thermal radiation to a receiver are separately 0 and 1 corresponding to different thermal radiation fluxes, then the risk of a tank or worker receiving a given thermal radiation flux can be determined by the thermal radiation flux equation.

In a crude oil tank farm containing more than one tank, the potential thermal radiation flux received by an object at an arbitrary location is the superposition of those from different tanks. For the optimization of space collocation and floor area of tank farm from inherently safe design, if the damage risk of an object from other overall tanks equals to the critical acceptable damage risk, and the corresponding floor area of tank farm is the minimum, it will certainly result in an optimal space collocation. The handling method for the mentioned problem inherently reduces property loss and casualty to some extent.

Introduction

In chemical industry, such as oil and petrochemical plants, quantities of hazardous materials that are produced, stored along with processes may lead to, under the hypothesis of ignition sources existence, explosions, fires and other serious accidents. Owing to the internal or external causes, the triggered explosions and fires may propagate affecting thus the reactors, tanks, pipelines and utilities erected in their vicinity as well as the on-site crew. This propagation may bring about calamitous consequences: failures of structure and human injuries or fatalities (Mebarki et al., 2014).

As an inherent way to reduce the foregoing propagation, a sound arrangement of process equipment and buildings should be determined and applied especially in the preliminary design phase. Some catastrophic incidents highlight this item. The explosions in Flixborough (1974) and Pasadena (1989) address attentions correlated with the layout issue. It is confirmed that improper distance between process equipment and occupied buildings (control rooms) is an effective factor contributing in the tragedies (Occupational Health and Safety Administration, 1990). Analogously, inadequate space between the isomerization process unit and the trailers is one of the major causes of accidents in Texas City refinery explosion (2005) (US Chemical Safety and Hazard Investigation Board, 2007). Additionally, it is also demonstrated by the calamities occurring in Seveso (1976) and Bhopal (1984) that a densely populated area in the vicinity of a hazardous plant can really be strongly affected (Joseph et al., 2005). These disasters mentioned above give a scenario of the equipment or units layout as a crucial factor deeply influencing the process safety.

Apart from the propagation of accidents, actually several other different issues incorporating constraints on process requirements, cost, safety, regulations, services and utilities availability have to be considered into the layout design (Tugnoli et al., 2013). It can be deduced that the optimal layout of vessels, tanks and pipes in a chemical plant is difficult and also a complicated and multidisciplinary task requiring the abundant knowledge from chemical engineering, physics and mathematics (mainly probability and optimization). The determination of optimal layout of units in a given place requires optimization calculation, namely a variety of distance constraints related to the process safety (Jung, 2016). Some values of separation distances for various specific equipment or units are customarily suggested and applied in the layout design (Mecklenburgh, 1985, Mannan, 2012). Other literature allowing for the cost aspects even safety issues in the optimal design of layout are represented (Nolan and Bradley, 1987, Penteado and Ciric, 1996, Patsiatzis et al., 2004, Díaz-Ovalle et al., 2010, Jung et al., 2010, Jung et al., 2011). Cozzani (Cozzani et al., 2006, Cozzani et al., 2007, Cozzani et al., 2009) and Gubinelli (Gubinelli et al., 2004) have deepened a more detailed safety analysis about the evaluation of appreciable accidental scenarios and consequence analysis involving the layout design. For anticipating safety issues related to the prevention of propagation in layout design, several indicators suitable in the early stages of layout design have been posed by Tugnoli (Tugnoli et al., 2007, Tugnoli et al., 2008a, Tugnoli et al., 2008b, Tugnoli et al., 2012).

Concerning the propagation of accidents and safety issues, in this paper, an integrated probabilistic framework (IPF) is introduced in order to deal with the optimal layout of equipment or units through its application in a case of crude oil tank farm. Generally, crude oil is stored in the external floating roof tank, and the potential initiating event due to the oxidative self-heating of pyrophoric iron sulfides on inside wall is explosion and then tank or bund fire (Mannan, 2012). It is worth then rigorously assessing the accident, i.e. the tank fire, in the present case. And the corresponding consequences focus on the major physical damage (thermal radiation) to the tanks and workers. Then a series of relevant procedures are developed to optimize the space collocation in a specific area based on the different acceptable thermal radiation separately corresponding to tanks and workers. In the end, the feasibility of these procedures is also discussed by a specific case study.

Section snippets

Primary tank fire and its occurrence probability

The iron sulfides which are formed on the inside wall and respiratory/safety valve inner cavity of tanks may generate, under given severe conditions, explosions and fires in the process of storage of crude oil. These initial accidents may propagate affecting the tanks, pipelines, power lines and facilities in their vicinity. Furthermore, this propagation may result in catastrophic consequences: structure failures that will lead to secondary sequences such as fires and explosions, as well as

Thermal radiation flux impacting on a receiver

A tank filled with high molecular weight-range hydrocarbon liquid is schematically shown in Fig. 1. It may be noted that on-site experience shows that in most cases of tank rupture, only a proportion of liquid in the tank is lost, and in many other cases, most of the liquid is retained in the tank. Therefore, it is commonly assumed that in terms of tank fire, the tank is effectively a pool of liquid with a flame burning on gas-liquid interface (Mannan, 2012). If the gas-liquid interface is

General procedures

According to Eq. (12), for a specified tank and with knowledge of its surrounding atmosphere conditions, the received thermal radiation flux for a target object is merely a function of the horizontal distance from the receiver to the corresponding nearest edge of fired tank. Increment of horizontal distance decreases the received thermal radiation flux to target objects. For a worker as the receiver, assuming the maximum thermal radiation flux for the human body is Er,p in a specific situation,

Conclusions

An integrated probabilistic framework (IPF) consisting a series of procedures to determine the optimal layout of tank farm is presented. For the probability of initial accident, it can be obtained by using the SVM method and Monte Carlo simulations. For the case study, the probability of tank fire due to the oxidative self-heating of iron sulfides is 1.854×106 per tank year. The main energy release of tank fire is thermal radiation, and the received thermal radiation flux of an object at X

Acknowledgement

The authors are grateful for the support given by National Key Research and Development Plan under Grant No. 2016YFC0800102, key project of National Natural Science Foundation of China under Grant No. 21436006, National Natural Science Foundation of China under Grant No. 51176070 and No. 21576136, PHC-CaiYuanpei (“Havu-Risk: Chemical industrial plants and domino effect: hazards, vulnerability, risks and sustainability” 32114 TE, 2014-2016), the Priority Academic Program Development of Jiangsu

References (42)

  • Q.B. Nguyen et al.

    Integrated probabilistic framework for domino effect and risk analysis

    Adv. Eng. Softw.

    (2009)
  • D.I. Patsiatzis et al.

    An MILP approach to safe process plant layout

    Chem. Eng. Res. Des.

    (2004)
  • A. Tugnoli et al.

    Safety assessment in plant layout design using indexing approach: implementing inherent safety perspective: Part 1-Guideword applicability and method description

    J. Hazard. Mater

    (2008)
  • A. Tugnoli et al.

    Safety assessment in plant layout design using indexing approach: implementing inherent safety perspective: Part 2-Domino hazard index and case study

    J. Hazard. Mater

    (2008)
  • A. Tugnoli et al.

    Supporting the selection of process and plant design options by inherent safety KPIs

    J. Loss Prev. Proc.

    (2012)
  • A. Tugnoli et al.

    Managing domino effects from a design-based viewpoint

  • R. Walker et al.

    Pyrophoric oxidation of iron sulphide

    Surf. Coat. Technol.

    (1988)
  • F.Q. Yang et al.

    Mechanism of mechanical activation for spontaneous combustion of sulfide minerals

    Trans. Nonferrous Met. Soc. China

    (2013)
  • C.L. Beyler

    Fire hazard calculations for large, open hydrocarbon fires

  • Center for Chemical Process Safety

    Guidelines for Chemical Process Quantitative Risk Analysis, Second Edition. American Institute of Chemical Engineers, New York

    (2000)
  • L. Gabriele et al.

    Hazard assessment of edible oil refining: formation of flammable mixtures in storage tanks

    J. Food Eng.

    (2011)
  • Cited by (0)

    1

    The authors contributed equally to this work.

    View full text