Showing papers in "Journal of Loss Prevention in The Process Industries in 1998"

Techniques and methodologies for risk analysis in chemical process industries

Abstract: This paper presents a state-of-art-review of the available techniques and methodologies for carrying out risk analysis in chemical process industries. It also presents a set of methodologies developed by the authors to conduct risk analysis effectively and optimally.

Inherently safer design based on rapid risk analysis

Abstract: The importance of inherently safer design (ISD) as a strategy to minimize risk of accidents in chemical process industries is being repeatedly stressed in recent years. The increasing number, frequency, and extents of damage caused by such accidents across the world have contributed to this thinking. However even as the need for ISD is being underscored, there are very few reports on precise methods to implement this concept. Significant recent reports are by Berge (1993 Berge, 1995) who has suggested a scenariobased design procedure in which construction of accident scenarios in a structured manner is made the basis of ISD. We have been developing and applying the concept of rapid risk analysis ( Khan & Abbasi, 1995 , Khan & Abbasi, 1996 , Khan & Abbasi, 1997a , Khan & Abbasi, 1997b , Khan & Abbasi, 1998a , Khan & Abbasi, 1998b ). In this paper we present an approach to ISD utilizing this concept. We believe, as detailed in this paper, that this approach is a significant improvement upon Berge's procedure in terms of ease, speed, and effectiveness.

Fusing ignition of various metal wires for explosion limits measurement of methane/air mixture

Abstract: In order to find the optimum condition for measuring the explosion limits of flammable gases by using metal wire fusing as the ignition source, the explosion limits of the methane/air system have been measured under various conditions in a bell jar type vessel of volume 3 dm 3 . Metal wires used were of copper, Nichrome, platinum, molybdenum and tungsten, with diameters ranging from 0.3 to 1.0 mm. Fusing was made by 12, 24, 36 and 48 V of direct current (DC), and also by 100 V of alternating current (AC). With these experiments, the effect of ignition abilities on explosion limits measurement has been made clear. As a result, it has been found that the fusing of metals of relatively high melting point, such as molybdenum, with a relatively high electric voltage, e.g., DC 36–48 V, is suitable for the explosion limit measurement. The explosion limits of methane in air have been determined to be 4.7–16.3%.

Reaction zone structures and propagation mechanisms of flames in stearic acid particle clouds

Abstract: Reaction zone structures and propagation mechanisms of two representative flames established in stearic acid (CH3(CH2)16CO2H) particle clouds have been investigated. The reacting zone structure was examined by using a micro-electrostatic probe and a high-speed schlieren system. A distinct difference was observed in the ion current fluctuations recorded across the two representative flames propagating through the clouds of the same total mass density of particles and different mass densities of the particles smaller than 60 μm in diameter. When the mass density of smaller particles was high, a single peak was recorded in the ion current fluctuation. On the other hand, when the mass density of smaller particles was low, multi-peaks of various heights and widths were recorded. In the former case, the single peak was considered to be attributable to a unitary and a relatively thin flame started burning in vapor generated by the evaporation of smaller particles in the preheat zone. The flame propagation mechanism in this case was inferred to be similar to that of a usual hydrocarbon–air premixed flame, although the reaction zone thickness is much larger than that of the premixed flame. In the latter case, the multi peaks of various shapes were considered to be attributable to strong combustion at blue spots far behind the schlieren front. The flame propagation in this case was inferred to be supported by the heat release due to combustion at the blue spots.

On the effect of venting large vessels with mass inert panels

Abstract: A numerical analysis indicated that in some situations venting by rather heavy venting devices such as panels or explosion doors will lead to substantial increase of the peak pressure in the considered enclosure compared to venting with burst disks or membranes of negligible mass. Vent sizing guidelines like the VDI 3673 consider a vent capability of explosion doors which has to be tested prior to their use. To study the effect of mass inertia of explosion doors or panels moved by translation, a set of experiments was carried out in a 50 m 3 silo with a height to diameter ratio H/D=4. The experiments showed, however, partially unexpected results. The peak pressure p red increased when changing membranes with inert devices (turbulent mixtures). When the mass of the devices was varied, no variation of p red was detected, though it could be demonstrated that the time necessary for full opening of the vent increased with higher mass. When venting with inert devices (quiescent gas mixtures at the instant of ignition), there are even situations which led to lower explosion pressures compared with membrane venting. Panels moved by translation are faster than explosion doors, but the flow through the vent seems to be less disturbed during venting in the case of explosion doors. The effect of pressure generation and flame front propagation during a vented explosion is dominated by conditions other than the mass of the devices, especially with the turbulence level of the mixture at the instant of ignition.

Kinetic system identification by using short-cut techniques in early safety assessment of chemical processes

Abstract: The chemical industry uses complicated reacting systems under a wide range of physical conditions. Besides production maximisation, stable and safe operation are important goals. This objective has to be considered in all the engineering calculations from the process design phase until the plant operation. Modelling and simulation of the processes accelerates and improves design. It may include qualitative information and/or heuristic rules. Process modelling, optimisation and control allow safe plant design and operation if enough information about the process is available. An experimental and a numerical technique will be combined in this paper. The `Differential Scanning Calorimetry' (DSC; Hohne, Hemminger & Flammersheim, 1996, Differential scanning calorimetry) is an experimental technique that can be used to investigate the mechanism and kinetics of a chemical process by measuring the thermal effect of the reaction following a very elaborated strategy. At early stages in process design, DSC is used as a screening tool to assess the thermal safety. Under the assumption of zero order Arrhenius kinetics, the activation energy and therefore the time to maximum rate under adiabatic conditions (TMRad) may be estimated (ANSI/ASTM, E 698-79; Keller, Stark, Fierz, Heinzle & Hungerbuhler, 1997, J. Loss Prev. Process Ind., 10, 31–41). A similar way can be applied to determine the kinetic constants and TMRad for an nth order reaction. Attempts to fit more complex kinetic models to DSC thermograms encounter many difficulties; one of them is model discrimination, a question not yet answered. The `Modified Integral Transformation Procedure' (MIP; Maria & Rippin, 1997 , Comp. & Chem. Eng., 21, 1169–1190) was proposed for quick process identification by considering previous information stored in data-banks and incomplete information about the process. The estimation technique is effective even if few but distributed process data are available and it can be easily coupled with other statistical data analysis and estimation techniques. The MIP is integrated in an expert system for process identification ( Maria & Rippin, 1996 , Comp. & Chem. Eng., S20, S587–S592) which facilitates computer-based plant analysis. It is the scope of this paper to investigate the effectiveness of using these coupled experimental and numerical short-cut techniques and an interactive data-bank in quick identification of complex chemical kinetics. If model discrimination is not possible, an experimental procedure to close existing data gaps most efficiently can be developed. The identified model can be further used in predicting optimum operating conditions for a chemical process by considering the desired product maximisation, waste and by-product minimisation, safety and operability as goals. The developed quick experimental and PC-coupled numerical identification and process analysis are exemplified in some simple complex kinetic cases. The effectiveness of the elaborated calculation methodology is discussed together with possibilities of further improvements.

Prediction of the maximum full lift safety valve two-phase flow capacity

Abstract: During an emergency blowdown of pressurized plant equipment, a discharge of a two-phase mixture across the safety valve, at least temporarily, can occur. Most calculation methods applicable for this flow condition do not contain geometrical data of the safety valve as parameters and are not yet, or only in small parameter ranges, validated. Therefore, experiments with commercially available full lift safety valves with flashing and non-flashing single- and two-component two-phase mixtures were carried out. The results of comparisons between the measured and the calculated mass fluxes demonstrate that merely with the methods according to the Isentropic Homogeneous Equilibrium Model, the ω-method of DIERS, the Homogeneous Frozen Flow by Nastoll (1985) and the procedure by Goslau and Weyl (1989) the measurements are qualitatively reproducible in limited parameter ranges. However, large deviations appear, particularly in the transition range to the single-phase flow of liquid and of vapour/gas, so that with no model an adequate prediction in the whole two-phase flow range is possible. By inserting a two-phase flow discharge coefficient, in analogy to the approach used in single-phase flow, the predictive accuracy of the Homogeneous Frozen Flow Model for non-flashing two-component flow is improved. In the case of flashing single-component flow, because of the boiling delay, the disequilibrium model of Henry and Fauske (1970) is extended by the two-phase flow discharge coefficient, and a better prediction accuracy is also obtained.

Dependence of flammability limits of a combustible particle cloud on particle diameter distribution

Abstract: Flammability limits and flame structure of a cloud of pure stearic acid (CH3(CH2)16COOH) particles have been investigated under a wide range of particle diameter distributions. The stearic acid in a reservoir was heated to become liquid of 74°C just above its melting point (68–71°C), and sprayed by a gas–liquid two phase nozzle. The liquid droplets turned into solid particles quickly during rising along a duct to the test section. Particle diameter distributions were controlled by widely changing spraying conditions. The structures and propagating velocities of particle cloud flames under various conditions were also examined by recording the direct emission from the flames, the images caused by laser light scattering, and the schlieren images of flames. Two flammability limits were found to exist. One is the lower flammability limit which mainly depends on the mass density of smaller particles (smaller than 60μm in diameter). The mass density of smaller particles at the lower flammability limit is about 3.0×10−5g/cm3. The other is the upper flammability limit which depends on the amount of all particles in the cloud. The mass density of all particles at upper flammability limit is about 34.0×10−5g/cm3.

Laser-based measurements in two-phase flashing propane jets. Part two: droplet size distribution

Abstract: This paper details the non-intrusive optical measurement of droplet size distribution in two-phase flashing propane jets, by the use of a diffraction-based technique. Centreline and lateral profiles of droplet size are presented. The data are compared with previously published data from similar flashing jets. The importance of careful and detailed data analysis is highlighted, to ensure that only valid data are obtained and reported, especially in the near-field regions of such jets. Even with considerable data analysis and evaluation, this technique may only be able to provide lower accuracy quantitative information and qualitative droplet size distribution trends.

Rocket missiles generated by failure of a high pressure liquid storage vessel

Abstract: A simple model has been developed for predicting the velocity of “rocket” missiles generated by failure at the lower end of a vertical cylindrical pressure vessel containing gas and a cold liquid. This model accurately predicts the velocities observed in a parallel experimental investigation. When modified to take account of liquid flashing to vapour during depressurization, the model has also been successfully applied to the case of “rockets” generated by failure of a vessel containing a hot liquid and a cover gas at a pressure sufficient to suppress boiling in normal operation. However, where “rockets” are generated by failure of a vessel containing a boiling liquid, the model significantly under predicts the “rocket” velocity in many cases. Here the enhanced experimental velocities are attributed to the increase in the apparent liquid volume, caused by the presence of the vapour bubbles, prolonging the liquid expulsion period. Simple correlations of the experimental data for all three cases have been developed using a parameter derived from the theoretical predictions. These define upper limit velocities for use in hazard assessment.

Evolved gases by simultaneous TG–MS technique and associated thermal hazard in drying of polyacrylonitrile

Abstract: Polyacrylonitrile (PAN) is an important industrial polymer. Its thermal decomposition is exothermic and the onset temperature is close to its drying temperature in industrial production. The thermal decomposition process of PAN is a self-heating type and can produce many toxic products. Thus there is a severe thermal hazard associated with its drying process. In the present study the products of thermal decomposition at various heating rates and isothermal temperatures in the presence and absence of oxygen are evaluated using simultaneous TG–MS technique to provide an insight into the mechanism of their liberation and to obtain qualitative knowledge of the associated hazards. The identified gaseous products liberate simultaneously in the temperature range 150–400°C and their evolution is dependent upon the heating rate and temperature. The principal reaction at lower heating rate is dehydrocyanation. Carbon dioxide liberation is observed, in addition to other products, in the presence of an oxygen-inert mixture atmosphere.

The role of turbulence in dust explosions

Abstract: The role of turbulence in dust explosions is effectively neglected by current vent sizing guidelines, which typically rely on experimental results obtained from tests reproducing near worst-case conditions. Work carried out at FMRC in a 64-m 3 enclosure has established the dependence of the pressure developed in vented dust explosions on the turbulence at the time of ignition. As a prelude to the use of these results in models of dust explosions, the generation of turbulence by jet flow into a closed volume has also been investigated. While detailed models will be necessary to advance the understanding of the fundamental processes of turbulent dust explosions, they will probably not be suitable for engineering calculations for some time, due to their complexity and high demand on resources. An alternative approach is proposed in the paper, involving the simulation of the flame propagation process by a lumped-parameter method. Encouraging results have been obtained with the standard k – e model (applied to the entire volume) in reproducing the global effects of the turbulence generation process due to jet flow injection. This treatment of the turbulence, when coupled with the flame propagation model already applied successfully to the simulation of turbulent gas explosions, is believed to offer a more appropriate solution for the performance of engineering calculations in practical systems, where the conditions of the process may only be defined in terms of overall quantities. The paper discusses the basis for the modelling approach and presents results obtained to date.

Low-temperature oxidation of coal at elevated pressures

Abstract: A mathematical model is presented to describe steady-state mass transfer and oxidation processes in coal at low temperatures in a fixed-bed flow reactor. The model incorporates the effects of partial pressure of oxygen, temperature and coal particle size, and accounts for the rate of coal oxidation and the composition of oxygenated products at high pressures. This is an important development since previous models did not include the effect of pressure in their formulation. It is found that, when the partial pressure of oxygen increases the rate of oxygen consumption increases accordingly. However, the influence of partial pressure of oxygen on the rate of oxidation is less pronounced when the pressure surpasses 1 MPa. In addition the model predicts that, for a constant partial pressure of oxygen, higher rates of oxygen consumption occur at lower total pressures. The same trends are also found for the concentration of oxygenated products at the reactor outlet. It is suggested that, the variation of partial pressure of oxygen leads to different concentration levels of oxygen at the surface and within the pores of coal particles, substantially affecting the rate of oxidation.

Design of a computer tool for the evaluation of the consequences of accidental natural gas releases in distribution pipes

Abstract: A computer system was designed to model releases of a gas to the atmosphere. This paper describes the mathematical models used to describe the phenomenon: source term (which can be applied to gases with an ideal behaviour), turbulent jet, jet fire (including thermal radiation vulnerability models) and atmospheric dispersion (both for light and neutral gases), as well as the software and hardware used in the system.

Simulation of VCEs by CFD modelling: an analysis of sensitivity

Abstract: The AutoReaGas code developed by TNO and Century Dynamics Ltd has been utilized to simulate the occurrence of a Vapour Cloud Explosion in a large fuel storage area. The sensitivity of the results to the main input variables has been investigated, together with the effect of fuel amount and fuel concentration inside the cloud. The marked sensitivity to several adjustable parameters, such as the turbulent modelling constant Ct, shows that the use of Computational Fluid Dynamics codes is not a straightforward task but still requires an adequate skill. From the whole set of computed results, some suggestions are derived, in order to improve the predictive power of the code.

Analysis of safety relief valve chatter induced by pressure waves in gas flow

Abstract: In this paper, on the basis of the allowable inlet line length according to the pressure surge and pressure loss criterion dedicated measurements of the valve opening behaviour are presented. Finally, the possibility for the prediction of valve disc oscillations based on numerical simulation will be discussed.

Laser-based measurements in two-phase flashing propane jets. Part one: velocity profiles

Abstract: This paper details the non-intrusive optical measurement of droplet velocity in two-phase flashing propane jets. Centreline and lateral profiles of axial velocity are presented and compared with a simple jet model. Non-intrusive techniques are presented as the only viable option for obtaining accurate data in such non-equilibriated environments. Even with these techniques, careful and detailed data analysis is required to ensure that only valid data is obtained and reported, especially in the near-field regions of such jets.

The chlorine release at Flix (Spain) on January 21st 1996: a case study

Abstract: In January 1996 a serious chlorine release took place at a plant for the production of chloromethanes in Spain, resulting in a toxic cloud that reached a neighbouring town. This work describes the cause of the accident and the ensuing sequence of events. Emergency procedures are also discussed. Finally, in the light of the conclusions reached some recommendations are drawn for industrial facilities handling chlorine.

The stability of toluene mononitration in reaction calorimeter reactor

Abstract: The kinetic parameters of toluene mononitration's reaction was measured using a reaction calorimeter (RC1) with an isothermal mode. These measured kinetic parameters were used to predict the theoretical transient temperature and fractional conversion simultaneously. The formulae for the evaluation of stability criterion in the toluene mononitration's reaction system in a non-isothermal batch reactor were derived from both transient mass and heat transfer equations. These stability-predicting formulae are explicitly expressed as a function of temperature, fractional conversion, physical properties and kinetic parameters. We then evaluate its criterion of stability using the derived equations and measured kinetic parameters.

Effects of fire on a container storage system—a case study

Abstract: The transportation of chemicals is often carried out using containers; a system characterised by the high density of the storage. In such cases, if there is a fire in one container it will affect the adjacent ones. The effects of such an accident are studied through the analysis of a case: a fire on two containers located on a ship. The possibility of BLEVE occurrence is analysed, as is the thermal radiation and the dispersion of smoke.

Risk analysis study of road transport of ethylene oxide

Abstract: A quantitative risk analysis is presented for road transportation of ethylene oxide. The procedure is outlined, discussing the amount and depth of information required. Based on analysis of the data relevant to accident frequency, consequences of historical accidents, meteorological conditions and resident population, the route was divided into a number of portions. Then, reference was made to a couple of weather conditions, assuming two typical release sizes and the complete series of possible outcome predictions for each release, which were estimated using a commercial software package for consequence analysis. All the information was combined to give, directly, the individual risk, as a function of the distance from the route, and the societal risk ( F – N curve), using a simple computer routine. The results showed that the examined transportation case represented a real hazard for the population and some mitigative actions were proposed, testing their effectiveness.

Hazard analysis and support tool for computer controlled processes

Abstract: Computer based control systems have been widely used in chemical processes with significant productivity improvements. However, the use of a computer introduces hazards as well as benefits. A hazard analysis framework for computer controlled processes is presented in this paper. The framework involves representation of the process, verification of the control logic, identification of safety critical events, and application of questions derived from industrial incident records. A computer tool is developed to support the hazard analysis framework. A novel modelling representation termed a Process Control Event Diagram (PCED) is described and used to express the control logic and its effect on the process. The control logic is verified by qualitative simulation based on the PCED. The whole procedure of applying the framework using the computer tool is illustrated by a case study.

Backpressure in a high-lift compensated pressure relief valve subject to single phase compressible flow

Abstract: Certified pressure relief valves may be used to prevent pneumatic explosion of pressurised systems. Previous published work has highlighted an allowable backpressure limit below which no effect on discharge capacity is noted, and a critical limit at which the valve becomes unstable. These limits are now identified with the normal operating characteristics of a valve (i.e. set pressure, lift, overpressure and blowdown). Tests have been performed on a valve designed to compensate for backpressure and flow characteristics are now presented. Critical backpressure ratio for choked compressible flow is identified, and the effectiveness of the compensation is investigated with respect to lift and discharge capacity. The fundamental physics governing the effect of backpressure is inferred from pressure measurements within the valve. Implications for further design work are stated.

Hazard review of a pesticides formulation plant and consequence assessment of accident scenarios in it

Abstract: In this paper a methodological approach is presented for risk and safety analysis in agrochemicals (pesticides) production plants. After a general study of the hazard sources in such a plant a list of possible accident scenarios was elaborated with emphasis on pesticides fires which result in the emission of highly toxic substances into the environment. These substances, under `favorable' atmospheric conditions, can be dispersed into the environment and constitute a significant hazard source both to the health of neighboring populations and, potentially, to the environment itself.

Gas explosion generated drag loads in offshore installations

Abstract: There is little guidance in place for secondary blast effects for offshore installations. In this paper, the origin of drag loading due to blast is explained, including the different nature of loading that could be expected for a range of explosion scenarios. The drag forces on pipe work, pipe supports grated decks and equipment are discussed. Methods on how to calculate these forces are described.

Fires, explosions and related incidents at work in Great Britain in 1994/95 and 1995/96

Abstract: Incidents in Great Britain reported to the Health and Safety Executive during 1994/95 and 1995/96 involving fires, explosions, runaway chemical reactions and unignited releases of flammable materials are reviewed. Statistical comparisons are made against previous years based on the materials involved, and a number of common themes and causes are identified.

Safety related questions for computer-controlled plants: derivation, organisation and application

Abstract: The use of computers in process control has improved productivity and product quality but has also caused a number of accidents. If we learn from these accidents we may be able to prevent them from happening again. This paper advocates a systematic approach to deriving and organising safety-related questions from past incidents and then applying the questions to consider the safety issues related to the whole life-cycle of computer-controlled plants. Over 170 questions were derived from 300 incident reports provided by two major organisations. The questions are organised into a structured framework so that relevant questions can be located easily when considering different aspects of a computer-controlled plant. Examples illustrating the application of the questions are given. The whole set of questions is listed in the Appendix.

Calculation of the stability criterion in the reaction of energetic materials

Abstract: The stability in the energetic materials' thermal decomposition reaction is developed from both transient mass and heat transfer equations simultaneously. The derived equation is the sufficient and necessary condition for the reaction system's stability, which was expressed as: φ (βθ+1) 2 (1−x) n nϕ(1−x) n−1 + exp −θ βθ+1 In this stability equation, a modified Semenov number φ is the expressed function of three kinetic parameters, i.e., n , ϕ and β and two variables, i.e., x and θ , respectively. A reaction system is stable if its evaluated modified Semenov number satisfies this stability criterion. In the case of the minimum modified Semenov number φ m , we can deduce the stability equation at critical condition as β 2 θ 2 c +2β 1+ ηβ n (1−x) θ c +1+ η n (1−x)(2β−1)=0 The criteria of stability for the explosives TNT, RDX, PETN and HMX were evaluated using this formula, which was expressed θ c as function of x at given n , η and various β as parameters.

Predicting damage of slender cylindrical steel shells under pressure wave load

Abstract: The present study deals with computer modelling of response of slender thin-walled steel shell structures, e.g. distillation columns and steel stacks under pressure wave load with the aim to increase the safety of the equipment mentioned. It is possible only on the basis of detailed knowledge of damaging mechanisms. It is also necessary to master a method of computer modelling and formulate assessment criteria to evaluate a degree of damage. The work generalizes the problem, which, in case of an unprotected structure, has already been discussed in the study ( Schneider, 1997 ). This study dealt at first with the problem of finding an external load which causes a limiting state of the structure, i.e. initiation of its collapse. This follow-up study is devoted, in the first place, to the generalization of the problem in a pressure–impulse diagram so that it could be possible, on the basis of known dimensional characteristics of the structure and known, possibly assumed, intensity of the explosion, to predict a degree of its damage and quantify, in that way, its survivability.

Gas explosion modelling of FPSO

Abstract: A Floating Production Storage Off-loading (FPSO) can be divided into four areas each with distinct characteristics of gas explosion development. The application of three generic types of gas explosion model is discussed. Suggestions are given for possible development of empirical and phenomenological models and the pitfalls associated with more flexible CFD codes are discussed. Experimental evidence and results of simulation show that there exists a critical cloud size, which we called runaway length, above which gas explosion could develop rapidly leading to damaging over-pressure. The success of control measures need to take account of the runaway length, e.g. an effective measure could include diverting a gas leak towards an area with a larger runaway length. The main structural surface on FPSO is the hull which tends to be strong. Drag loading on pipework, equipment and grated decks could be more significant than simple over-pressure load.