Process and Plant Safety

© Springer-Verlag GmbH Germany, part of Springer Nature 2020

U. HauptmannsProcess and Plant Safetyhttps://doi.org/10.1007/978-3-662-61484-6_1

1. Introduction

Ulrich Hauptmanns¹  

(1)

Schönebeck, Germany

Ulrich Hauptmanns

Email: ulrich.hauptmanns@ovgu.de

Whoever demands absolute safety, ignores the law of life.

Keywords

Process IndustrySafety SystemSafety TechnologyLatin WordCleaning Work

1.1 Introduction

The production of the process industry¹ often involves hazards. Their nature can be both physical and chemical. Physical hazards derive from operating conditions that may be extreme, such as very low or very high temperatures and pressures. Chemical hazards are those associated with the materials present in the process, which can be toxic, flammable, explosible, or release energy due to spontaneous² reactions. Indeed, it is the necessity to put the substances into a reactive state in order to enable one to produce the desired products that may lead to hazards.

A further complication stems from the fact that some of the properties of the substances can vary with changes of process parameters such as temperatures, pressures or concentrations, or that these changes may give rise to or favour unwanted side reactions, as was the case in the Seveso accident, where larger quantities of dioxin than usual were generated and released to the environment (cf. [1]).

In addition, dangerous properties, if not present under nominal operating conditions, may evolve upon contact of process media with auxiliary media such as coolants or lubricants. After release, reactions with substances present in the environment, e.g. the humidity of the air, may give rise to dangerous properties.

Nevertheless a concretization of the hazard potential is normally not to be expected, since the design, construction, erection, and operation of the plants are based on the state of technology, respectively safety technology³ (cf. [2]). Hence, they are supported by a broad base of experience, which, depending on the country, is reflected by the respective laws, rules, and regulations. A good overview of this topic is provided by the Guideline Plant Safety [3].

According to [3] the design of a plant has to be such that the containment of hazardous substances inside the plant, i.e. vessels, pipework, reactors etc. is ensured. This does not only result in demands on the mechanical resistance of the components of the plant, but requires safety systems to be introduced, which in case of undesired loads (mostly excessive temperatures and/or pressures) are to guarantee the integrity of the containment by pressure relief, emergency trips, emergency cooling etc.

If all components were to function with perfection and, in addition, the measures of safety management were perfect plants would be absolutely safe.

This is, however, not the case and cannot be achieved. Apart from the—although remote—possibility of wrong dimensioning (e.g. walls too weak) components of engineered systems can fail, humans can commit errors in operating the engineered system or external threats such as flood, storm or lightening may lead to failures within the plant. Thus, temperature and pressure increases or other damaging events may be triggered. In addition, it is conceivable that safety systems are not available due to component failures. Probabilities for such events may be assessed. However, the instant in time of a component failure, human error or destructive external event cannot be predicted.

Hence, despite careful design, construction and operation of plants accidents cannot totally be avoided. Whatever may happen will happen with a certain probability. Therefore the probability of an accident⁴ can only be reduced by appropriate measures. To achieve this is the objective of risk management.

Yet, a risk remains, i.e. a probability (or more precisely an expected frequency) that a damage of a certain type and impact occurs. In a process plant this may be a fire, an explosion or a toxic release, which may affect both humans and the environment. It is the price to be paid for the desired product. The damage can affect employees, the population at large or both, as becomes evident from Table 1.1.

Table 1.1

Some accidents in the process industry [9]

The protection of the employees is ensured by a number of laws, regulations and guidelines (cf. [5, 6]). The justified interest of the population in safety, the protection according to the Federal Pollution Control Act (BImSchG) [7], is guaranteed by the licensing procedure.

Two fundamental approaches in licensing are conceivable:

(1)

the license is granted solely on the basis of fulfilling the above mentioned requirements; risk is not assessed.

(2)

In addition to (1) statements on risk have to be made and certain risk criteria to be met.

The procedure according to (1) is used in the Federal Republic of Germany and that of (2), for example, in the Netherlands.

It has to be emphasized that the operating systems of a plant are dimensioned by the same procedure with both approaches. Requirements for the systems are specified, for example, the quantity of heat to be extracted from a reactor for an exothermic reaction. The corresponding calculations are performed using mathematical models reflecting the underlying laws of nature. Results in this case may be, for example, the power of the coolant pump, the necessary surface for heat transfer, or the pipe diameters. This procedure is called deterministic.

The safety design of a plant results from extensive analyses (cf. [2]) to be discussed later. The dimensioning of safety systems is also carried out deterministically. It is based on the concept of disturbances that have to be avoided,⁵ for example a cooling failure in a reactor for an exothermic reaction. This is the basis for determining the type and capacity of the safety system coping with it. Its quality and degree of redundancy may then be determined

(1)

by indeterminate legal terms in regulations (cf. [4]) such as reliable measuring device or

(2)

probabilistically⁶ based on risk criteria.

As mentioned before, the approach according to (1) is that used so far in Germany. However, in the meantime probabilistic requirements for safety systems are derived from risk considerations in fulfilment of the standards on functional safety [10–12]. This corresponds to (2).

There is a recent tendency to measure the safety achievements by indicators (so-called key performance indicators) (cf. [13, 14]). These refer on the one hand to past performance (lagging indicators) and on the other to future performance (leading indicators).

In order to give an impression of standards achieved in the German process industry the following assessment is made. The accident statistics [15] shows that there was no fatal accident involving members of the public during 10 years of operation of the 7800 plants subject to the Major Accident Ordinance [14]. On this basis a Bayesian zero-event statistics leads to a coarse assessment of 6.4 × 10−6 a−1 for a fatality outside a plant (vid. Example 9.4).

Figure 1.1 provides an impression of the safety performance concerning labour accidents comparing the chemical industry with figures for the industry at large.

Fig. 1.1

Fatal accident rate FAR (fatalities per 10⁸ working hours) for the chemical industry and the industry in general in Germany [16]

Plant and process safety encompasses all the areas required for designing and building a process plant and implementing the corresponding processes (amongst them process, mechanical, and civil engineering). As a rule time-dependent processes have to be treated, since we are usually concerned with deviations from nominal operating conditions. The latter are considered as safe if a rigorous implementation of safety has accompanied the design and erection of a plant and is a permanent concern during its operation. The compliance with these assumptions should, of course, be checked in the context of a safety analysis.

Safety deals with stochastic events, for example the moment of occurrence of an accident, and stochastic boundary conditions (e.g. the weather at that moment). These together with lacks of knowledge about some of the phenomena to be described and imperfections in models and input data lead to uncertainties, which are normally compensated by safety factors and often lead to procedures based on conventions.

The treatment of uncertainties has substantially progressed in recent years (cf. [17–27]). However, their detailed theoretical treatment is beyond the scope of the present text, so that only procedures with particular relevance for practical applications are explained.

In what follows the physical and chemical phenomena causing the hazard potential of process plants are treated in Chaps. 2 and 3. Chapters 4, 5, 6 und 7 are dedicated to engineered and organizational measures that are devised to avoid that the hazard potential harms employees and the public at large. Chapters 8, 9 und 10 deal with the determination of engineering risks. In this context the methods of plant system analysis and models for assessing accident consequences are presented. They serve to identify hazard potentials and to develop concepts for coping with them. Hence, they influence the safety design of plants and their safe operation. An important aspect of the safe design of plants is the concept of functional safety, which is treated in Chap. 11. Finally, Chap. 12 is devoted to the determination of appropriate safety distances between industrial installations and the surrounding population, which may be an additional safeguard for reducing the consequences of an accident.

References

1.

Mannan S (ed) (2005) Lees’ loss prevention in the process industries, hazard identification, assessment and control, 3rd edn. Elsevier, Amsterdam

2.

SFK (2002) Störfallkommission beim Bundesminister für Umwelt, Naturschutz und Reaktorsicherheit (Hrsg.), Schritte zur Ermittlung des Standes der Sicherheitstechnik, SFK-GS-33, Januar 2002

3.

SFK (1995) Störfallkommission beim Bundesminister für Umwelt, Naturschutz und Reaktorsicherheit (Hrsg.): Leitfaden Anlagensicherheit, SFK-GS-06, November 1995

4.

Zwölfte Verordnung zur Durchführung des Bundes-Immissionsschutzgesetzes (Störfall-Verordnung – 12. BImSchV), Störfall-Verordnung in der Fassung der Bekanntmachung vom 15. März 2017 (BGBl. I S. 483), die zuletzt durch Artikel 1a der Verordnung vom 8. Dezember 2017 (BGBl. I S. 3882) geändert worden ist, Stand: Neugefasst durch Bek. v. 15.3.2017 I 483; Berichtigung vom 2.10.2017 I 3527 ist berücksichtigt, Stand: Zuletzt geändert durch Art. 1a V v. 8.12.2017 I 3882 (German implementation of the DIRECTIVE 2012/18/EU OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL of 4 July 2012 on the control of major-accident hazards involving dangerous substances, amending and subsequently repealing Council Directive 96/82/EC/, Seveso III-Directive)

5.

Verordnung über Sicherheit und Gesundheitsschutz bei der Verwendung von Arbeitsmitteln (Betriebssicherheitsverordnung -BetrSichV), Betriebssicherheitsverordnung vom 3. Februar 2015 (BGBl. I S. 49), die zuletzt durch Artikel 1 der Verordnungvom 30. April 2019 (BGBl. I S. 554) geändert worden ist

6.

Gesetz über die Bereitstellung von Produkten auf dem Markt (Produktsicherheitsgesetz –ProdSG),Produktsicherheitsgesetz vom 8. November 2011 (BGBl. I S. 2178, 2179; 2012 I S. 131), das durch Artikel 435 der Verordnung vom 31. August 2015 (BGBl. I S. 1474) geändert worden ist

7.

Gesetz zum Schutz vor schädlichen Umwelteinwirkungen durch Luftverunreinigungen, Geräusche, Erschütterungen und ähnliche Vorgänge (Bundes-Immissionsschutzgesetz – BImSchG), Bundes-Immissionsschutzgesetz in der Fassung der Bekanntmachung vom 17. Mai 2013 (BGBl. I S. 1274), das zuletzt durch Artikel 1 des Gesetzes vom 8. April 2019 (BGBl. I S. 432) geändert worden ist (Immission Act)

8.

StörfallVwV—Erste Allgemeine Verwaltungsvorschrift zur Störfall-Verordnung vom 20. September 1993 (GMBl. S. 582, ber. GMBl. 1994 S. 820)

9.

http://​www.​aria.​developpement-durable.​gouv.​fr/​

10.

Functional safety – Safety instrumented systems for the process industry sector – Part 1: Framework, definitions, system, hardware and application programming Requirements (IEC 61511-1:2016 + COR1:2016 + A1:2017); German version EN 61511-1:2017 + A1:2017

11.

DIN EN 61511-2:2019-02;VDE 0810-2:2019-02, Functional safety – Safety instrumented systems for the process industry sector – Part 2: Guidelines for the application of IEC 61511-1 (IEC 61511-2:2016); German version EN 61511-2:2017

12.

DIN EN 61511-3:2019-02;VDE 0810-3:2019-02, Functional safety – Safety instrumented systems for the process industry sector – Part 3: Guidance for the determination of the required safety integrity levels (IEC 61511-3:2016); German version EN 61511-3:2017

13.

Guidance on SAFETY PERFORMANCE INDICATORS—Guidance for Industry, Public Authorities and Communities for developing SPI Programmes related to Chemical Accident Prevention, Preparedness and Response, (Interim Publication scheduled to be tested in 2003–2004 and revised in 2005), OECD Environment, Health and Safety Publications, Series on Chemical Accidents, No. 11

14.

Sugden C, Birkbeck D, Gadd S Major hazards industry performance indicators scoping study, HSL/2007/31

15.

https://​www.​infosis.​uba.​de/​index.​php/​de/​zema/​index.​html

16.

Lipka B (2009) Deutsche Gesetzliche Unfallversicherung (DGUV), personal communication October 2009

17.

Morgan GM, Henrion M (1990) Uncertainty—a guide to dealing with uncertainty in quantitative risk and policy analysis. Cambridge University Press, New YorkCrossref

18.

Balakrishnan S, Georgopoulos P, Banerjee I, Ierapetriou M (2002) Uncertainty considerations for describing complex reaction systems. AIChE J 48(12):2875–2889Crossref

19.

Watanabe N, Nishimura Y, Matsubara M (1973) Optimal design of chemical processes involving parameter uncertainty. Chem Eng Sci 28:905–913Crossref

20.

Nishida N, Ichikawa A, Tazaki E (1974) Synthesis of optimal process systems with uncertainty. Ind Eng Chem Process Des Dev 13:209–214Crossref

21.

Knetsch T, Hauptmanns U (2005) Integration of stochastic effects and data uncertainties into the design of process equipment. Risk Anal 25(1):189–198Crossref

22.

Hauptmanns U (1997) Uncertainty and the calculation of safety-related parameters for chemical reactions. J Loss Prev Process Ind 10(4):243–247Crossref

23.

Hauptmanns U (2007) Boundary conditions for developing a safety concept for an exothermal reaction. J Hazard Mater 148:144–150Crossref

24.

Reagan MT, Naim HN, Pébay PP, Knio OM, Ghanem RG (2005) Quantifying uncertainty in chemical systems modelling. Int J Chem Kinet 37(6):368–382Crossref

25.

Reagan MT, Naim HN, Debusschere BJ, Le Maître OP, Knio OM, Ghanem RG (2004) Spectral stochastic uncertainty quantification in chemical systems. Combust Theory Model 8(3):607–632Crossref

26.

Hauptmanns U (2008) Comparative assessment of the dynamic behaviour of an exothermal chemical reaction including data uncertainties. Chem Eng J 140:278–286Crossref

27.

Hauptmanns U (2012) Do we really want to calculate the wrong problem as exactly as possible? The relevance of initial and boundary conditions in treating the consequences of accidents. In: Schmidt J (ed) Safety technology—applying computational fluid dynamics. Wiley-VCH, Weinheim

Footnotes

1

The term process industry comprises firms from the chemical, petrochemical, pharmaceutical and food industries as well as the production of steel, cement and the like.

2

Without apparent reason from the Latin word sponte from its own accord.

3

State of safety technology: the state of development of advanced processes, installations and procedures that permit one to take for granted the practical aptitude of a measure for avoiding accidents or limiting their consequences. When determining the state of safety technology comparable processes, installations and procedures have to be considered that have been successfully applied in practice [4] (translated by the author).

4

Accident: an event such as an emission, a fire or an explosion of major impact that leads to a disturbance of the specified operation* in a site or a plant subject to this ordinance (Author’s remark: this refers to the Major Accident Ordinance [4]) that leads immediately or at a later stage to a serious hazard or material damage within or outside the site involving one or several hazardous substances as listed in annex VI part 1 para I no. 4.

*Specified operation is the operation for which a plant is designed and appropriate. Operating regimes not covered by the valid license, posterior impositions or applicable legal requirements do not belong to the specified operation. The specified operation comprises the

normal operation including necessary human interventions such as the taking of samples and including the storage with filling, transfer and refilling procedures,

plant commissioning and its start-up and shut-down,

trial operation,

maintenance, inspection, repair and cleaning work as well as

periods of temporary stand-still [8] (translated by the author).

5

In the field of nuclear engineering this is referred to as design-basis accident.

6

Based on probability considerations derived from the Latin word probabilis: assumable, likely, credible.

© Springer-Verlag GmbH Germany, part of Springer Nature 2020

U. HauptmannsProcess and Plant Safetyhttps://doi.org/10.1007/978-3-662-61484-6_2

2. Hazardous Properties of Materials

Ulrich Hauptmanns¹  

(1)

Schönebeck, Germany

Ulrich Hauptmanns

Email: ulrich.hauptmanns@ovgu.de

Keywords

Flame FrontFlame SpeedLaminar Burning VelocityDust ExplosionAdiabatic Flame Temperature

2.1 Flammability

A large number of the materials handled in the process industry are flammable. They react with oxygen releasing thermal energy. In general the oxygen stems from the air but other oxidants have to be considered as well, for example hydrogen peroxide or ammonium nitrate that easily release oxygen. Furthermore, substances like chlorine or fluorine can play the role of an oxidant.

In general combustion takes place if a flammable material enters into contact with an energy source, e.g. an electrical spark or a hot surface, and thus receives energy. If solid or liquid materials are concerned their temperature has to be raised first to such an extent that vapour is produced by vaporization or disintegration. These vapours can form flammable mixtures with air just as flammable gases. If the energy supply is sufficient a self-sustaining exothermic reaction occurs.

The conditions for a combustion process are shown in Fig. 2.1. It presents the so-called fire triangle, which comprises the necessary elements of a combustion process, namely fuel, oxidant and energy.

Fig. 2.1

Fire triangle

The consequence of a combustion process is either a fire or an explosion. Which of the possibilities occurs depends on the boundary conditions to be treated below. In general the approach is empirical. For example conditional probabilities (the condition is the preceding release) of 0.6 for a fire and 0.4 for an explosion after the release of a flammable gas or liquid are given in [1].

The safe handling of flammable materials requires the knowledge of their properties, which are normally described by safety parameters. These parameters are not, as a rule, constants of nature but values that are determined under fixed boundary conditions. This leads to the use of standardized measuring apparatuses (vid. [2–4]). When employing these parameters to judge real situations an eye must therefore be kept on the prevailing boundary conditions.

Example 2.1

Empirical frequencies for fires and explosions

The ARIA-database indicates the following numbers of events as a consequence of hydrocarbon releases: a = 1,748 events explosion or fire, b = 656 events explosion and c = 1,554 events fire.

Determine the conditional probabilities (the condition is the release whose probability of occurrence is assumed here to be equal to 1) for the different events.

Solution

The sum of the numbers of events with fires and explosions amounts to

However, this includes events where fire and explosion occurred jointly. Their number is

From this we have

events with an explosion only and

events with a fire only.

Hence we obtain the following conditional probabilities:

Only fire: 1,092/1,748 = 0.625

Only explosion: 194/1,748 = 0.111

Fire and explosion: 462/1,748 = 0.264

If the explosion is considered to be the dominating event and the probability for fire and explosion is added to the probability for only explosion the result is in good agreement with that of [1]. □

2.1.1 Safety Parameters for Flammable Gases and Vapours

2.1.1.1 Explosion Limits

Combustion can occur only if the mixture of fuel and oxygen lies within a certain range. This is described by the lower and upper explosion limits (LEL and UEL). In older references theses limits are referred to as the lower and upper limits of flammability (LFL and UFL) (vid. [4]). They represent the volume ratio¹ of fuel vapour in air. Below the lower explosion limit the mixture is too lean, above the upper limit it is too rich for combustion to occur. The explosion limits are not fixed values. They depend on whether we deal with a mixture with air or with oxygen. Furthermore they are influenced by (vid. [4, 5]):

pressure,

temperature,

direction of flame propagation,

type and location of the source of ignition, in particular ignition energy,

type and size of the space (closed, open, geometry),

possibly the amount of inert gas in the mixture,

flow regime of the gas,

gravitational field.

Additionally they depend, as already mentioned, on the boundary conditions of their measurement, as illustrated by Table 2.1. In general the most flammable mixture is close to but not exactly equal to the stoichiometric one [5].

Table 2.1

Upper and lower explosion limits according to different sources

The explosion limits may be calculated approximately by (vid. [6])

(2.1)

(2.2)

In Eqs. (2.1) and (2.2) cst is the stoichiometric concentration (volume percent of fuel in air). In case of a stoichiometric equation of combustion of the form

(2.3)

we have

(2.4)

and hence

(2.5)

However, Example 2.2 shows that the differences between calculated and measured values are considerable. Hence, whenever possible measured values are to be used.

This applies as well for the pressure dependence of the explosion limits. The following logarithmic relationship is given for the pressure dependence of the UEL (vid. [6])

(2.6)

In Eq. (2.6) p denotes the absolute pressure in MPa. The equation does not represent the measured values, as is evident from Table 2.2. The values for 1 bar agree because they are introduced into the equation as the reference value

.

Table 2.2

Dependence of the explosion limits on initial pressure (measured values from [4], calculated values (bold print) according to Eq. (2.6))

asince 100% is the maximum, the value is merely a formal result

bmeasured at an initial pressure of 50 bar

According to [4] the lower explosion limit decreases slightly with increasing initial pressure whilst the upper limit increases strongly. Exceptions from this rule are the gases hydrogen and carbon monoxide. The lower explosion limit of hydrogen at first rises slightly with increasing initial pressure and then decreases with further pressure increase. In the case of carbon monoxide the range between the explosion limits narrows at first with increasing initial pressure and remains constant with a further increase.

With an increase in temperature the range between the lower and upper explosion limits widens for all flammable gases. The relative change of the lower and upper limits is similar for many flammable gases. Hence, it may well be approximated by the following linear relationship

(2.7)

In Eq. (2.7) xB(T) denotes the volume ratio of the gas at temperature T and xB(T0) that at the reference temperature T0, e.g. ambient temperature. The positive sign applies to the upper explosion limit, the negative sign to the lower limit (vid. [4]). Factors for K are given in Table 2.3, where KL applies to the lower limit and KU to the upper.

Table 2.3

Temperature coefficients KL and KU for selected flammable gases (vid. [9])

*Calculated from experimental data for use in Eq. (2.7)

aTemperatures up to 400 °C

bTemperatures up to 250 °C

The above considerations apply to a mixture of a single flammable gas and air. If several gases, e.g. I, are involved that do not react with one another, the principle of Le Chatelier is invoked and we obtain

(2.8)

(2.9)

In Eqs. (2.8) and (2.9) yi is the molar fraction of material i in the total mixture; LELi and UELi are the corresponding explosion limits.

Experience tells that this estimate agrees fairly well with the measured values of the lower explosion limit for similar flammable gases. The upper limit shows larger deviations. The equations should be applied with care to safety technological questions, since the deviations may lie on both the safe and the unsafe side [4].

Example 2.2

Uncertainties of the explosion limits taking propane as an example

The explosion limits of a material depend on numerous boundary conditions. Hence different measurements result in different values as shown in what follows taking the lower explosion limit of propane as an example. The following values in volumetric percent are given

xn: 1.7; 2.1; 2.2; 2.1; 2.1; 1.7; 2.1.

Let us assume they represent N = 7 independent measurements (independence does often not apply since values from the same source are quoted in several references). Then the explosion limit may be assumed to be a random variable, i.e. a variable that adopts certain values with certain probabilities. Random variables are described by probability distributions (vid. Appendix C). In what follows the logarithmic normal (lognormal) distribution (vid. Sect. 9.​3.​4) is used to represent the values

As mean value of the logarithms of the values of xn we have

and as the corresponding standard deviation

The pertinent probability distribution and probability density function, simply termed probability and probability density or pdf, are represented by Fig. 2.2.

Fig. 2.2

Probability and probability density of the lower explosion limit of propane

The percentiles are to be interpreted such that the corresponding percentage of the lower explosion limit lies below the respective percentile value. □

Example 2.3

Determination of the lower and upper explosion limits

Determine the lower and upper explosion limits of acetylene, hydrogen and ammonia for a pressure of 1 bar.

Solution

The solution is based on Eqs. (2.1) to (2.5). The results are compiled in Table 2.4.

Table 2.4

Calculation of the lower and upper explosion limits for several materials

Comparison with the measured values from Table 2.1 shows that the results are merely approximations. This underlines that it is necessary from a safety point of view to use measured values. □

Example 2.4

Temperature dependence of explosion limits

The lower and upper explosion limits of methane are to be determined for the temperatures 100, 200, 300 and 400 °C.

Solution

Combination of Eq. (2.7) with Table 2.3 leads to the results of Table 2.5. They are in good agreement with the measured values, as is demonstrated in Fig. 2.3.

Table 2.5

Temperature dependence of the explosion limits of methane

Fig. 2.3

Comparison of the temperature dependence of measured and calculated explosion limits of methane

□

Example 2.5

Calculation of the lower and upper explosion limits of natural and petroleum gas

Natural and petroleum gas have the main components given in Table 2.6.

Table 2.6

Composition of natural and petroleum gas in mol%

Determine their lower and upper explosion limits.

Solution

According to Eqs. (2.8) and (2.9) we obtain

for natural gas and

for petroleum gas.

2.1.1.2 Explosion Limits for Mixtures

Mixtures of flammable gases and oxidant were treated in the preceding Section. In practice often mixtures have to be assessed that in addition contain an inert gas. The corresponding situation is represented by Fig. 2.4. It is subsequently described following [9].

Fig. 2.4

Explosion region of a flammable gas presented in triangular coordinates [9]

The explosion limits form a boundary line enclosing all flammable compositions; the explosion range contains all flammable mixtures. Points on the sides of the triangle represent two-component systems because the fraction of the third component is zero there. Pure substances are represented by the edges of the triangle. The fractions of the remaining two components are equal to zero there. In Fig. 2.4 the upper edge represents the pure flammable gas, the lower right edge the pure inert gas and the lower left edge the pure oxidant. If a certain amount of flammable gas, inert gas or oxidant is added to the mixture A, a new mixture results. If the component is added continuously, the point A moves along a straight line in the direction of one of the edges of the diagram (denoted by the arrows in Fig. 2.4). For example, if a certain quantity of flammable gas is added, a new composition according to point B is obtained after the mixture has homogenized.

2.1.1.3 Ignition Temperature

According to [7] the ignition temperature of a flammable gas or flammable liquid is determined in a standardized experimental set-up (vid. [10]). It is the lowest temperature (in °C) of the heated glass bulb, on whose concave wall the inhomogeneous gas air or vapour air mixture of the examined material (at a pressure of 1,013 mbar) is just ignited showing flames (readily ignitable mixture). Hence, it constitutes an appropriate measure for the propensity of materials to be ignited on hot surfaces. This enables one, amongst others, to assign materials to temperature classes according to safety technological criteria.

It must be emphasized that we deal with a measurement that requires no further energy source in order to produce an ignition. Table 2.7 gives ignition temperatures for selected materials.

Table 2.7

Ignition temperatures (from [7])

aValue for coarse orientation

2.1.1.4 Minimum Ignition Energy

The minimum ignition energy (MIE) is a parameter for judging the incendivity by important sources of ignition such as electrostatic discharge and mechanical spark. It represents the smallest possible amount of energy capable of just igniting the most flammable gas/air or vapour/air mixture in such a way that a flame occurs that is not restricted to the immediate vicinity of the igniting spark. The value of the MIE depends on both the testing apparatus and the testing procedure. It is determined on the basis of the energy of the discharge spark of a capacitor that is applied to the most flammable mixture under standard conditions (20 °C und 1,013 mbar) [11]. The most flammable mixture is considered to lie in the range of 0.9–1.4 (according to other sources 0.8–2) times the stoichiometric mixture (vid. [3]). The latter can be calculated according to Eq. (2.5).

The minimum ignition energy is determined according to

(2.10)

In Eq. (2.10) E is the ignition energy in J, C the capacitance of the capacitor in Farad and U the voltage applied to the capacitor in V. By varying the energy E the energy amount is identified that is just sufficient to ignite the mixture under examination, the MIE.

If the ignition source is not at rest relative to the surrounding mixture, for example in the case of a flowing medium, heat is lost and the MIE increases [6]. Table 2.8 contains values of the MIE for selected materials.

Table 2.8

Minimum ignition energies (MIEs) for normally ignitable materials (standard conditions) (from [3])

Example 2.6

Ignition of hydrogen

A capacitor with a capacitance of 560 pF (1pF = 10−12 F) is charged with a definite current of U0 = 220 V. Would its discharge ignite hydrogen?

Solution

describes the time-dependent voltage in the capacitor, U(t), and τ0 the time constant of the charging system. For simplicity’s sake the asymptotic voltage of the capacitor is used. It amounts to U0 and is obtained in theory for t → ∞ and in practice after a period of time of approximately five times the time constant.

Inserting the numerical values in Eq. (2.10) we obtain

Since 0.0136 mJ > 0.012 mJ (lower limit of the interval indicated in Table 2.8) the cautious analyst should expect ignition to occur.

2.1.1.5 Burning Velocity

According to [11] the burning velocity is the movement of the flame front in a homogeneous gas/air mixture per unit of time at a right angle with the flame front into the unburnt mixture. The burning velocity is determined by heat conduction, diffusion and the flow process, with the latter resulting from the expansion of the combustion gases. It depends on the initial temperature, the amount of oxygen introduced, the degree of mixture and catalytic effects (e.g. traces of steam, smoke or dust). The burning velocity is measured with respect to the unburnt gas. Hence, it differs from flame speed that is the velocity of the flame front with respect to a fixed observer. The flame speed is one or two orders of magnitude greater than the laminar burning velocity because of the acceleration produced by the expanding combustion products.

The burning velocity is usually determined on pre-mixed flames from a Bunsen burner in laminar flow regime (see Sect. 2.1.1.7). It is then called laminar burning velocity. In case of turbulent flow the burning velocity is many times the laminar burning velocity and does not depend on the properties of the mixture alone. Within the explosion limits the burning velocity is an appropriate parameter for describing flame propagation. Burning velocities depend on pressure and temperature [5]. Table 2.9 presents laminar burning velocities for selected substances. In some cases these velocities may be represented by polynomials, such as

(2.11)

for liquefied petroleum gas (LPG) [13] with the main components 27.65 vol% propane and 68.28 vol% butane and

(2.12)

for natural gas [14].

Table 2.9

Laminar burning velocities for selected materials

In Eqs. (2.11) and (2.12) vburn is the burning velocity in cms−1 and

the ratio of the molar stoichiometric requirement of air, nL,min, and the available number of moles of air ( $$ \upphi = 1 $$ , stoichiometric). This value is called equivalence ratio and is the reciprocal of the air-to-fuel ratio $$ \uplambda $$ .

Example 2.7

Determination of the burning velocities for petroleum gas and natural gas

Determine the burning velocity of petroleum gas and natural gas for different equivalence ratios in steps of 0.1.

Solution

Application of Eqs. Gl. (2.11) and (2.12) leads to the values of Table 2.10; they are shown in Fig. 2.5.

Table 2.10

Laminar burning velocities for petroleum gas and natural gas as a function of the mixing ratio $$ \upphi$$

Fig. 2.5

Laminar burning velocities for petroleum gas and natural gas as a function of the equivalence ratio

2.1.1.6 Critical Slot Width and Maximum Experimental Safe Gap

Propagation of flames is hindered if they have to cross a small slot. This phenomenon is characterized by the critical slot width and the maximum experimental safe gap (MESG). According to [3] the critical slot width is the width of a slot with given length that after an explosion of the readily ignitable mixture or flammable vapour just prevents the ignition of the mixture on the other side of the slot. The critical slot decouples the space in which an explosion occurs from the surrounding flammable atmosphere. The MESG is the lowest value of the critical slot widths. It is measured by varying the composition of the mixture [7]. Details on the measuring process can be found in [3].

The most flammable concentration is found to lie between 0.9 and 1.4 respectively 0.8 and 2 times the stoichiometric concentration. The latter may be determined from Eq. (2.5). Table 2.11 provides MESGs for selected materials.

Table 2.11

Maximum experimental safe gaps (MESGs) for selected flammable materials (from [7])

Example 2.8

Determination of most easily ignitable concentrations

Determine the most easily ignitable concentrations for the hydrocarbons from Table 2.8 and for hydrogen assuming that it occurs at 1.1 times the stoichiometric composition.

Solution

The calculations are based on Eqs. (2.3) to (2.5). The results are shown in Table 2.12.

Table 2.12

Assessment of the most easily ignitable concentrations

□

2.1.1.7 Basic Flame Types

After presenting several of the safety parameters of fire and explosion protection different types of flames are briefly discussed here. The presentation largely draws upon [15].

Basically we distinguish between pre-mixed and non pre-mixed flames (formerly called diffusion flames). With pre-mixed flames the mixing between fuel and oxidant occurs before combustion with non pre-mixed flames mixing and combustion are simultaneous. A pre-mixed flame is obtained, for example, if the air supply of a Bunsen burner is opened; if it is closed the flame becomes non pre-mixed. Another example of a non pre-mixed flame is a burning candle.

Further differentiations are found in Fig. 2.6. In what follows they are briefly commented upon.

Fig. 2.6

Differentiation of flame types (according to [15])

For pre-mixed flames the velocity of combustion is limited by the kinetics of the combustion process. In case of a laminar non pre-mixed flame the limitation usually stems from the diffusion velocity of air into the fuel, with turbulent non pre-mixed flames on the other hand the kinetics becomes more determining.

Laminar pre-mixed flame

The combustion velocity of a freely burning flat flame into the unburnt mixture can be described by the laminar burning velocity (vid. Sect. 2.1.1.5). In doing this different regimes of combustion can be distinguished on the basis of the equivalence ratio $$ \upphi$$ .

$$ \upphi$$  < 1 lean (there is oxygen left after combustion)

$$ \upphi$$  = 1 stoichiometric (only combustion products remain after the combustion)

$$ \upphi$$  > 1 rich (fuel is left after the combustion)

If the velocity of the unburnt gas is smaller than the laminar burning velocity the flame flashes back into the outlet opening. In the opposite case blow-off occurs (slight separation from the outlet) and at even higher flow velocities the flame lifts.

Turbulent pre-mixed flame

The transition from laminar to turbulent flames occurs for Re ≈ 2,000 with the Reynolds number referring to the flame. It is smaller than in the unburnt mixture because the viscosity of gases rises with increasing flame temperature. The combustion process of a turbulent pre-mixed flame can be controlled well. However, for safety reasons it is not readily applied because flammable mixtures may accumulate and hence explode.

Laminar non pre-mixed flame

Non pre-mixed flames are characterized by more complex chemical processes than pre-mixed ones and may comprise the entire spectrum 0 <  $$ \upphi$$  < ∞. They occur if a pure fuel flows from an outlet opening and is then mixed with the surrounding air and thus with oxygen by diffusion and entrainment. Contrary to pre-mixed flames non pre-mixed flames do not propagate and hence cannot be characterized by the laminar burning velocity. For the flame length we have [16]

(2.13)

and according to Jost (cit. in [5])

(2.14)

and according to [7]

(2.15)

In Eqs. (2.13) to (2.15) L is the flame length in m, c0 the concentration of the fuel at the outlet opening (fuel/air ratio: usually equal to 1, respectively 100%), cst is the corresponding stoichiometric concentration, $$ {\dot{\text{V}}} $$ the volumetric flow rate in m³s−1 and D the diffusion coefficient in m²s−1.

The equations suggest that the flame length at constant mixture ratio only depends on the volumetric flow rate, i.e. it is independent of the cross-section of the outlet opening. With a given outlet cross-section it is approximately proportional to the flow velocity. The differences between the relationships point to modelling uncertainties.

Turbulent non-premixed flame

With increasing velocity at the outlet the laminar flame becomes turbulent. The transition between the two regimes occurs at Re ≈ 2,000. Contrary to laminar non pre-mixed flames its length does not depend on the velocity at the outlet.

Example 2.9

Determination of the lengths of non pre-mixed flames

Determine the flame lengths of non pre-mixed laminar flames for the flammable gases hydrogen, carbon monoxide and acetylene at a volumetric flow of 0.0001 m³s−1.

Data:

Solution

The bases are Eqs. (2.13) and (2.5) as well as Eqs. (2.14) and (2.15). The resulting flame lengths are contained in Table 2.13.

Table 2.13

Flame lengths for selected gases in m

□

2.1.1.8 Adiabatic Flame Temperature

The adiabatic flame temperature indicates the thermal power radiated by the flame. It represents an upper limit,

because it is determined assuming combustion without losses and

because often the ionization and dissociation of the combustion products, which start above 1,370 °C and consume energy thus reducing the temperature, are not accounted for [5].

The adiabatic flame temperature applies to pre-mixed flames, usually in stoichiometric mixtures with oxygen from air. Non pre-mixed flames reach temperatures of appr. 1,500 K. If the oxidant is pure oxygen instead of oxygen from the air the occurring temperatures are 700–800 K higher.

The first law of thermodynamics applies to an adiabatic system (heat losses dq = 0) at constant pressure [17]. Accordingly the sum of the enthalpies is equal to zero. Hence, we obtain for calculating the adiabatic flame temperature

(2.16)

In Eq. (2.16) ni is the number of moles of input material number i with the corresponding difference of enthalpy between its temperature (normally ambient) and the standard temperature (298.15 K), $$ \Delta {\text{h}}_{{{\text{in}},{\text{i}}}} $$ . It is equal to zero if its initial temperature equals the standard temperature, mj is the number of moles of combustion product j and $$ \Delta {\text{h}}_{{{\text{out}},{\text{j}}}} $$ the corresponding difference in enthalpy between the initial temperature and the adiabatic flame temperature of the combustion products; ΔHc is the enthalpy of combustion of the fuel in question. For ΔHc the negative of the net calorific value of the fuel is used, i.e. ΔHc = −Hu.

The enthalpy at the standard pressure of 101.3 kPa and at a temperature of t in K/1,000, h⁰ in kJ/mol, can be obtained from [18], where the following representation is used for a number of materials

(2.17)

In Eq. (2.17) $$ {\text{h}}_{298.15}^{0} $$ is the standard enthalpy (at 101.3 kPa and 298.15 K), which cancels on forming the enthalpy differences in Eq. (2.16). The relationship between enthalpy and heat capacity at constant pressure is useful for practical applications

(2.18)

where T = t·1000 is the temperature in K.

Table 2.14 contains coefficients for applying Eq. (2.17) and Table 2.15 experimentally and theoretically determined adiabatic flame temperatures for selected materials. The latter were calculated iteratively from Eq. (2.16). For this purpose the following reaction equations were used assuming air to consist of oxygen and nitrogen only so that 1 mol of oxygen is accompanied by 3.774 mol of N2:

Table 2.14

Coefficients for determining the enthalpy of selected materials (from [18])

Table 2.15

Experimental and theoretical values of the adiabatic flame temperatures for selected materials (initial temperature 25 °C)

aSiegel and Howell (cited according to [5])

bLewis and von Elbe (cited according to [5])

Acetylene

(2.19)

Ethanol

(2.20)

Ethylene

(2.21)

Methane

(2.22)

Hydrogen

(2.23)

Example 2.10

Determination of the adiabatic flame temperature of acetylene

Determine the adiabatic flame temperature of acetylene for an initial temperature of 25 °C using the following mean values for the heat capacities at constant pressure. The net calorific value of acetylene, Hu, is taken from Table 2.15.

Solution

Based on Eq. (2.16) and accounting for Eq. (2.18) we obtain for the reaction Eq. (2.19)

C2H2 + 2.5 O2 + 9.435 N2 → 2 CO2 + H2O + 9.435 N2

the following enthalpy balance

Solving this equation for T gives an adiabatic flame temperature of 3,056.8 K, where the difference with Tab. 2.15 is explained by the use of mean values. □

2.1.1.9 Explosions

As already mentioned the ignition of a flammable material may cause a fire or an explosion. An explosion is understood to be the sudden and violent release of energy. The violence of the explosion then depends on the rate of energy release. For example, the energy stored in a car tyre causes an explosion if the tyre bursts. On the other hand it may be gradually dissipated through a puncture. The following types of energy may lead to an explosion:

physical energy

pressure energy of gases

strain energy of metals

electrical energy

heat energy

quick phase change (steam explosion)

chemical energy

combustion reactions

dust explosions

runawayreactions

decomposition

polymerization

explosives

nuclear energy (uncontrolled chain reaction)

In what follows explosions caused by combustion reactions are treated first. Runaway reactions, decomposition and polymerization are a subject of Chap. 3, explosives are treated in Sect. 2.5 and explosions resulting from physical energy in Sects. 10.​6 to 10.​9. Nuclear explosions are outside the scope of this book.

Two types of explosions due to chemical energy are distinguished

deflagration

and

detonation.

In a deflagration the flammable mixture burns relatively slowly. Flame propagation is mainly determined by molecular diffusion and turbulent transport processes. Mixtures of hydrocarbons and air burn in the absence of turbulence, i.e. under laminar or almost laminar conditions, with flame speeds of the order of 5–30 m/s. If there is no confinement this is too slow to produce tangible overpressures and only a flash fire is produced. That is why there is always turbulence involved in a vapour cloud explosion (turbulent flame speeds 100–300 m/s), which increases the rate of combustion and hence the overpressure [12].

A detonation is totally different. The flame front moves as a pressure wave closely followed by a combustion wave, which supplies the energy nourishing the pressure wave. A stationary detonation pressure wave reaches the velocity of sound corresponding to the hot gases. This velocity lies well above that for unburnt gases. For hydrogen-air mixtures the velocity is of the order of 2,000–3,000 m/s compared with 300 m/s in air of 0 °C.

A detonation causes a stronger pressure wave and more destruction than a deflagration. Whilst the peak pressure of a deflagration produced by a mixture of hydrocarbons and air in a confined space amounts to about 8 bar, 15–20 bar are reached following a detonation. Contrary to a deflagration a detonation does not have to occur in a confined space in order to produce such high pressures [5].

A detonation is not a stable but a fluctuating process. This finds its expression in a cellular structure of shock and reaction waves. The cell structure depends on the type of fuel and the composition of the mixture. More reactive mixtures have smaller cell sizes. Hence, cell size is a measure for the propensity of a material to detonate (cf. Table 2.16). Table 2.16 shows as well that lower ignition energies are required for a deflagration, whilst the direct triggering of a detonation requires high energies [19]. The range of concentrations in which a detonation is possible is always smaller than that for ignition (UEL–LEL) (cf. [4]). However, it becomes broader with increasing initial pressures and temperatures.

Table 2.16

Characteristic detonation cell sizes and ignition energies for the deflagration and detonation of selected stoichiometric fuel-air mixtures (from [19])

A deflagration may turn into a detonation, especially if it propagates through a pipe. Such a process is called Deflagration-Detonation-Transition (DDT). Research has not yet totally clarified its characteristics (cf. [20, 21]).

Basically it may be stated that turbulence enhancing circumstances such as obstacles, building structures, and confinements as well as high ignition energies favour the transition from deflagration to detonation. The same is true for high initial pressures and temperatures.

One must differentiate as well between confined and unconfined explosions. Typical confinements are vessels and pipework as well as buildings. Unconfined explosions (outdoors) exhibit other characteristics than confined ones (vid. Sect. 10.​6.​3).

2.1.1.10 Maximum Pressure and Maximum Rate of Pressure Rise

Deflagration

The strength of an explosion is characterized by its maximum pressure and its maximum rate of pressure rise. The standardized methods for measuring these parameters are described in [22].

The maxima of the explosion parameters depend on the vessel volume. Whilst the maximum pressures of conventional fuels (flammable vapours) may generally be approximately constant and only dependent on vessel geometry the maximum rate of pressure rise may adopt very different values depending on the type of fuel and vessel volume.

The volume dependence of the maximum rate of pressure rise of a flammable gas or vapour can be described by the cube root law

(2.24)

In Eq. (2.24) the unit of the vessel volume has to be compatible with that of KG, i.e. m³ for the KG values from Table 2.17, which additionally also contains maximum pressures; dp/dtmax then results in bar s−1.

Table 2.17

Characteristic values for the explosion of flammable gases and vapours (5 l sphere for explosion tests, ignition energy E = 10 mJ, standard conditions) from [3]

aExtrapolated value

Example 2.11

Maximum rates of pressure rise for the deflagration of methane and hydrogen

Determine the maximum rates of pressure rise for vessel sizes between 5 l and 10 l in steps of 1 l for methane and hydrogen.

Solution

Using Eq. (2.24) in conjunction with Table 2.17 we obtain

for methane and

for hydrogen

The results for the maximum rates of pressure rise in bar s−1 are given in Table 2.18 and in Fig. 2.7 for an even larger range of volumes.

Table 2.18

Maximum rates of pressure rise for deflagrations of methane and hydrogen in bar s−1

Fig. 2.7

Maximum rates of pressure rise for methane and hydrogen as a function of vessel volume

Detonation

□

A detonation is usually modelled as a one-dimensional shock wave (vid. [5, 23, 24]). For this purpose a coordinate system is used that moves with the combustion front (velocity Vs = −w1 in the coordinate system of an outside observer, also called laboratory system, vid. Figure 2.8). The following relations are used in the model in order to relate the state after the detonation (subscript 2) with that before the detonation (subscript 1):

Fig. 2.8

Schematic for deriving the equations governing a detonation

Conservation of mass flux

(2.25)

Conservation of momentum

(2.26)

Conservation of energy

(2.27)

Inserting Eqs. (2.25) and (2.26) in Eq. (2.27) we have

(2.28)

Additionally the ideal gas equation of state is invoked

(2.29)

Equation (2.28) establishes a relationship between pressure and specific volume (enthalpy is related via heat capacity and Eq. (2.29) with pressure and specific volume). It is called Hugoniot equation or Hugoniot shock adiabatic and consists of thermodynamic quantities only.

The foregoing equations use the following nomenclature:

w

: velocity of gases relative to the combustion front in m/s

p

: pressure in Pa

v

: specific volume in m³/kg

h

: enthalpy in J/kg

Hu

: net calorific value (heat of reaction) in J/kg

M

: molar mass of the flammable gas

Rm

: universal gas constant Rm = 8.3145 (J/mol K)

R

: mass basis gas constant R = Rm /M in J/(kg K)