Archive for the ‘ATEX Directive’ category

Intrinsic safety in hydrogen/oxygen mixtures

July 9, 2008

Occasionally it is necessary to make measurements in hydrogen/oxygen mixtures. The increased use of fuel cells and the use of hydrogen as a vehicle fuel have increased the frequency with which this requirement occurs. The requirement occurs usually within process vessels since when the mixture is released the problem becomes a mixture of hydrogen/oxygen /air which is a slightly different problem. Conventional intrinsically safe equipment certified to the IEC standard is considered to be adequately safe under normal atmospheric conditions. The ATEX guidelines on the 94/9/EC directive state ‘A product within a potentially explosive mixture without the presence of air is not in the scope of the directive’. It follows therefore that that a safety analysis of an electrical installation in a hydrogen/oxygen mixture must be based on a risk analysis using the best available information. The conventional IS certification ensures a satisfactory level of construction but additional consideration must be given to the levels of voltage, current and power used. This note proposes an approach, which the author considers, achieves an acceptable level of safety.

Available data
Most of the available data is quite old, and this note relies heavily on an SMRE paper P6 published in 1974. [I would appreciate copies of any recent data relating to hydrogen/oxygen mixtures that anyone has and if necessary I can then amend this note.] The IEC apparatus standard IEC60079-11 includes some oxygen-hydrogen test mixtures, which simulate the 1.5 safety factor for the various gas groups. For example the IIC mixture is 60% hydrogen and 40% oxygen. The most readily ignited mixture of hydrogen and oxygen is said to be 33% hydrogen and 67% oxygen and quite how this compares with the sensitivity of the IIC test mixture is not clear from the available evidence. The ignition energy of the most sensitive mixture, when tested using a break-flash tester appears to be approximately 11mJ with a threshold voltage of 10V. In the literature there are very low figures for capacitive derived ignition energy [2mJ] but these are obtained using high voltages and very special electrodes, which are not relevant to intrinsically safe circuits.

A practical approach is that where possible the circuit voltage should be kept below the threshold voltage, since this appears to be well established and also to limit the inductive energy. Figures of 6V and 6mJ would achieve an acceptable level of safety and are not impractical. The 6mJ has a factor of safety of three on the 18mJ permitted in ‘ia’ circuits. [40mJ with a 2.25 factor of safety]. The permitted inductance or L/R ratio in a hydrogen/oxygen mixture can therefore be derived by dividing the figures for IIC by three. Frequently the L/R ratio is the predominant factor. At 6V capacative energy is not a problem but some large capacitors have a significant self-inductance, which can create a problem. Setting a limit of 10mF is reasonable and in line with current practice.

There is very little information on the effect of oxygen on the ignition temperature of hydrogen [5600C]. Probably the ignition temperature is much lower and the use of T4 [1350C] apparatus seems a prudent and practicable solution. The 1,3W small component relaxation for T4 apparatus is possibly not applicable to a hydrogen/oxygen mixture hence a reduction by a factor of two seems a reasonable precautionary measure.

The proposed limits are therefore an ‘ia’ IIC system with T4 apparatus, a circuit voltage of 6V, an inductance or L/R ratio derived by dividing the IIC figures by three and a capacitance of 10mF.  Where the small component relaxation is used for temperature classification then the matched power should be reduced by a factor of two. With the factors of safety implicit in these values it seems acceptable to use ‘simple apparatus’ with the usual parameters.

Practical implications
There are commercially available galvanic isolator interfaces for the majority of the sensors used in this type of Zone 0. For example suitable interfaces are available for switches, proximity detectors, thermocouples and RTDs. Some forms of oxygen content monitors also meet these criteria.

The range of shunt diode safety barriers, which satisfy the 6V requirement, is limited. The usual solution is the two channel 3V 10W barrier which when combined with ‘simple apparatus’ would have its IIC cable parameters of 100mF, 130mH and 69mH/W reduced to 10mF, 43mH and 23mH/W. The matched power of 450mW is acceptable and the permitted L/R ratio makes the installation a practical possibility.

Many of the field mounted transmitters which would not normally be mounted directly in this atmosphere but having sensors in the Zone 0 have sensor input parameters which satisfy the proposed requirements of this note.

Using conventional intrinsically safe instrumentation in a hydrogen/oxygen atmosphere with the proposed limitations can be considered to be acceptably safe. The resultant installation is not intrinsically safe in accordance with the IEC standards but achieves a comparable level of safety.
The proposed approach is arguably the only acceptable approach using the conventional methods of explosion protection.


Cable parameters and all that

June 12, 2008

The intrinsically safe [IS] system standard [IEC60079-25] discusses in detail how to draw up the system documentation and in particular how to calculate the permitted cable parameters. Unfortunately the standard has to take into account all the possible variations and hence the process looks quite complicated. In the majority of applications the simple precaution of using a single source of power with no significant capacitance or inductance across its output terminals [Ci &Li] and field devices with only small input capacitance and inductance [Ci & Li] removes any problems. The permitted cable parameters [Cc,Lc & Lc/Rc] are then the specified output parameters [Co,Lo & Lo/Ro] of the source of power and there is no difficulty. The majority of certified apparatus has low values of Ci and Li and hence in simple systems this simplification is applicable. For this purpose negligible values of Ci and Li are 5nF and 10mH respectively.

The experimental work done by PTB to support the FISCO standard suggests that adding conventional cable to an IS system makes it safer. It would require extensive further testing [possibly reduced by the use of the PTB spark test simulator] to extend the principle established in the FISCO standard to include all other forms of IS system. Unfortunately there is no finance available to do this work. If anyone can raise the funds, I am sure a means of getting the work done could be found. Until this optimum solution is proven and becomes part of the standard it remains necessary to create system documentation which defines the permitted cable parameters.

Frequently installations do not have a problem because the permitted values are higher than can be created by the cables used. The most common limitation is imposed by the permitted capacitance of the higher voltage circuits. The usual limitation is that the permitted capacitance for 28V circuits in IIC for ‘ia’ and ‘ib’ circuits is 83 nF. The IEC code of practice [IEC 60079-14] suggests practical maximum cable parameters of 200pf/m, 1mH/m and 30 mH/W, which suggests a limit on cable length of 400m. A 400m cable can be considered as having an inductance of 400mH which corresponds to a current of approximately 300mA which is not common in IS circuits. Where high currents do occur then the permitted L/R ratio is usually in excess of the 30 mH/W limit and there is no problem If the cable lengths on an installation are less than 400m and the systems comply with the above restrictions then quite a lot of repetitive documentation can be avoided by including in the safety documentation a statement such as ‘The installation does not use cables longer than 400m and the permitted cable parameters are not greater than 200pF/m, 1mH/m and 30 mH/W and consequently no further consideration of cable parameters is recorded.’

Similarly if an ‘ia’ or ‘ib’ system is used in a Zone 2 location then it can usually become an ‘ic’ system and the permitted inductance increases by a factor of 2.25. The permitted capacitance increases by a variable factor which reduces as voltage increases but at 28V the permitted capacitance becomes 272nF and the cable length problem is effectively removed. A general statement can be utilised, such as ‘Where the installation uses ‘ia’ or ‘ib’ systems and the locations only require ‘ic’ systems the need to consider cable parameters is for all practical purposes removed and consequently no further consideration of cable parameters is recorded.’

An even more obvious case for not considering cable parameters is when a IIC system is used where a IIB gas classification is applicable. The permitted inductance is multiplied by 4 and the capacitance by a large variable factor. At 28V the capacitance moves from 83 to 650nF. The argument for not being concerned about cable parameters becomes very powerful


If possible avoid using sources of power or IS apparatus with significant values of Ci and Li. If you are using cable lengths less than 400m there is no real problem. If you are using ‘ia’ or ‘ib’ systems where an ‘ic’ system is acceptable then there is no significant problem. If you are using a IIC system where a IIB system is acceptable then there is no problem.

If in any analysis of an intrinsically safe system the solution shows that there is a significant cable parameter problem then recheck the calculation. Real problems do occasionally occur but there are always very unusual factors.

Finally if you think you have made a mistake sleep easy because the probability is that the cable made it safer anyway

Intrinsically safe multicores

January 22, 2008

It is an opportune time to review the use of multicores containing more than one IS circuit because of the introduction of the ‘ic’ concept. [For the uninitiated ‘ic’ is intrinsic safety without faults intended for Zone2 use and will replace the ‘nL’ concept of the type ‘n’ standard in time] A further factor is that the subject has now been introduced into the draft documents of the system standard [IEC 60079-25] whereas previously guidance was confined to the code of practice [IEC 60079-14]. The reason for this is that the information is required for Group I [mining] applications and IEC 60079-14 is only applicable to surface industry. The alignment of the two documents is being carefully monitored, in the vain hope of avoiding conflicting requirements.

As an aside the CDV [voting document] of the next edition of the system standard should emerge in the near future. If you are involved in IS installations you should obtain a copy from your local standards body and make your opinions known since this is the last opportunity to influence the content of the document.


There are a number of basic requirements for IS cables. At first this seems strange since the open circuit and short circuit of field wiring is taken into account in the design of an IS system. The requirements are intended to reduce the probability of multiple faults such as random earth faults to an acceptable level. It is necessary to establish the inductive and capacitive parameters of a cable and this is difficult unless the cable is of a known conventional construction. There is a need for all cables to be suitable for their intended environment. Resistance to chemicals and extremes of temperature is a frequent requirement. In practice the requirements have not been challenged because they align with the requirements for a reasonably operationally secure installation. The origins of many of the requirements in the IS standards are a pragmatic mixture of safety and operational needs.


The basic requirements of an IS cable is that the insulation shall withstand a 500V insulation test and that the minimum strand diameter should be 0,1 mm. The latter requirement is concerned with all the current flowing through a single strand causing hot wire ignition. There is little or no justification for this requirement but it has been there since 1960 and is impossible to dislodge


Multicore cables have additional requirements of a minimum insulation thickness of 0,2 mm and a modified insulation test. The sensible approach is to use the two types of multicore, which are considered to prevent the interconnection of the different, IS circuits within the multicore. The most easily defended multicore is a Type A cable in which each IS circuit is contained within an individual screen. There are then two screens between the circuits and the probability of an interconnecting fault penetrating both screens is considered to be acceptably low. Where the cable can be well supported and protected against mechanical damage a cable without interposing screens, known as a Type B cable can be used. The conventional multicore between the IS interface and the distribution junction box protected by a cable tray is an example of this type of installation. There is no requirement to protect against the rampant fork lift truck so beloved by standards writers, a modicum of common sense on the level of protection should be applied.

It is permitted to use a multicore, known as a Type C cable, without screens or mechanical protection but it is then required to consider faults between the IS circuits [up to four open circuits and two short circuits]. This analysis is difficult if all the circuits are resistive limited, and if one or more of the circuits is non-linear then the analysis is even more complex. All the circuits within this type of cable take the gas classification and the level protection of the least well protected circuit. For example if one of the circuits is ‘ic’ then all the circuits become ‘ic’. This type of cable is permitted but is best avoided. There are occasions when several IS circuits have to be taken via the same route and continuously flexed, such as the monitoring circuits on a robot arm. In these circumstances using separate cables within an hydraulic tube rather than a multicore has much to commend it.

It is necessary to establish the cable parameters of the multicore. Where Type A or B cables are used and the cores used for a particular circuit can be readily identified then the parameters of that combination can be identified and used to determine the safety of the system. It is therefore desirable to choose a cable with readily identifiable combinations of cores [fortunately this is usually possible] If this is not possible then the parameters of the worst possible combinations of cores must be used. It is usually cable capacitance, which becomes a problem. If a Type C cable is used then the worst possible combination must be used, which is another good reason for not using this type of cable.


Whenever possible avoid the use of a Type C cable. It is permitted but the safety analysis is quite complex and the probability of fully complying with all the requirements low. A well-supported Type B cable is usually the most economic solution. There is a small question of cross-talk between signals if high frequency signals are used. [This is a practical problem, not a safety problem] The use of separately screened circuits as required by Type A cables is technically commendable but may not be practicable for availability and economic reasons.

The use of mechanical tools in hazardous areas

December 16, 2007

Purpose of note
From time to time concern is expressed about the possible risk of frictional sparking caused by the use of steel tools in hazardous areas. This note attempts to put this risk in perspective and make a positive proposal on acceptable practice.

It is recognised that frictional sparking between certain materials can ignite a flammable gas with only a very small amount of mechanical energy. The classic example is the early cigarette lighter, which with a flick of the thumb ignited petroleum vapour by rotating a serrated steel wheel against a flint. The materials which cause a lot of concern are light alloys because heating small particles of these materials causes a thermal reaction which produces a higher than expected temperature. The use of such alloys in electrical equipment for use in hazardous areas is discouraged in the General Requirements [IEC 60079-0] which imposes limits on the use of aluminium, magnesium, titanium and zirconium in Zone 0 and 1. Interestingly there are no restrictions on the use of these alloys for Zone 2 applications except for fans and ventilating screens, presumably it is accepted that the risk of frictional sparking causing ignition is low enough for it to be ignored.

Most of the available technical evidence supports the view that ignition by small hot particles is related to spark ignition energy and not to ignition by hot surfaces. That is the IIC gases [hydrogen, acetylene and carbon disulfide] are easier to ignite than IIB and IIA gases. The minimum level of impact required to produce ignition is difficult to establish but a glancing blow is thought to be most incendive. A recent analysis of the available literature plus some experimental evidence carried out by a German Test House suggests that a steel upon steel impact has to be greater than 3Nm for ignition to occur. At 10Nm there is a 10% risk of igniting a IIC gas and a very low probability of igniting a IIB gas. [3Nm is created by a 1Kgm weight falling through 30cm]. These are significant levels of impact and are not likely to occur unintentionally in the maintenance and inspection of instrument systems.

In the conditions where maintenance is normally performed hydrogen is unlikely to achieve the ideal concentration for ignition [20% by volume in air]. Hydrogen has very low density and low viscosity and hence disperses rapidly on release in a non-enclosed location. It has to achieve a concentration in excess of 10% before its ignition energy is less than that of ethylene, hence in the majority of Zone 1 locations where servicing of instrumentation is carried out, it becomes effectively a IIB gas.

The use of tools
The use of those tools normally used in servicing instruments is not likely to produce an impact which will could create a frictional spark capable of igniting any gas/air mixture. The weight of such tools as screwdrivers and keys to socket headed screws [Allen keys] are such that they would have to achieve a significant velocity before producing an impact of 3Nm. The use of a correctly chosen socket spanner is a safer practice than the use of non-steel alternatives which are frequently unsatisfactory because they are not sufficiently robust. It is advisable to avoid the use of rusty tools, but this is normal practise. This manually operated type of tool is normally considered capable of causing a single spark. Activities, which cause multiple sparks such as grinding and sawing, do require special consideration, since they pose a greater risk.

There are obvious areas of risk, normally outside the activities of the majority of instrument technicians such as the use of a hammer and cold chisel to remove rust from an enclosure or the dismantling of aluminium scaffolding , both of which require some thought.

Guidance within IEC standards on the use of tools is not available. Several petroleum industry codes do give guidance and they all suggest that there is no significant risk. A CENELEC standard EN 1127-1 has an informative Annex A which takes an extremely cautious and impracticable view on the subject. It proposes that the use of steel tools should not be permitted in a Zone 1 without a gas clearance certificate where the gas producing the risk is a IIC gas. Possibly this prohibition can be justified if the tools involved are heavy and of the kind which might be necessary for work on major pieces of mechanical equipment. A blanket approach, which includes the tools normally, used by an instrument engineer is not justifiable.

EN 1127-1 adds a number of IIB gases to the prohibition. These are hydrogen sulfide, ethylene oxide and carbon monoxide. Why these particular gases are singled out is not apparent and it would be interesting to find the basis of this selection.

The use of steel tools of the type normally used to maintain instruments does not cause an unacceptable risk of ignition due to frictional sparking in Zones1 and 2 whatever the gas/air mixture creating the possible hazard.

There is a need for anyone working in hazardous areas to be aware of the risk associated with frictional sparking, but the careful use of steel tools is acceptable

In anticipation of a possible comment, I realise that gases are not classified and that it is equipment, which is classified for use with certain gases. However if this note was written using phrases such as a ‘gas which requires the use of IIC equipment’ it becomes even less comprehensible. This justifies the use of ‘IIC gases’ even though it is not pedantically correct.

The future role of the humble paper clip in Hazardous Areas

December 6, 2007

Apart from their traditional use of holding the multiple sheets of paper that one is required to have with them when performing inspections on equipment in hazardous areas, what are the other uses for the humble paper clip these days ?

The first one that came to mind was a piece of test equipment for IS switch circuits. Chris Towle has always suggested the use of a straightened out paper clip to fault find cabling or other faults on switches connected to IS interfaces. Acting as a simple shunt and being classed as Simple Apparatus, it can be used at convenient points along the loop starting at the Hazardous Area terminals of the interface to see at what point the problem occurs.
A second use came to an end this week with the release of the OSHA Directive on Dusts (link). Previously NFPA654 (Standard for the Prevention of Fire and Dust Explosions from the Manufacturing, Processing, and Handling of Combustible Particulate Solids ) has used 1/16″ or the thickness of our favourite tool, as being the sufficent depth of dust required on a piece of equipment for it to be considered to be posing a significant risk of flame spread or secondary explosions. The recent OSHA document has now reduced this to be 1/32″ based on recent tests. This is of course a rule of thumb and the actual thickness will depend on other factors but it has reduced the applications for our paper clip. We therefore need you to come up with other uses, suggestions please via the comments section.


PS The good news is that the OSHA Directive does suggest the use of Intrinsic Safety as the protection technique to be used for instrumentation in areas where there are dust hazards so not everything is that bad. Perhaps this will finally increase the use of the technique in North America

Earthing –bonding and all that

September 4, 2007

The perennial problem of earthing and bonding is raised from time to time. The basic principles that are applicable to all bonding problems is that it is potential difference that causes problems and explosions and that all currents should be provided with a well defined low impedance return path. Within Europe there is a strong movement to train cows to stand on one leg during thunderstorms so that they avoid the potential difference which could ignite the methane in their stomachs. These learned ‘certified ‘ beasts will be branded with the appropriate ATEX marking.


A recent situation where these principles were applied was to consider the bonding of an IS installation on an off-shore rig which was subject to lightning induced surges, the primary concern being the bonding of the screen. The installation, which emerged, is illustrated in this diagram



NOTE. The numbers shown on the diagram and used throughout the text are indicative of the order of values commonly encountered. They are intended to assist in understanding the subject and are not intended to be precise or applicable in all circumstances.

Almost all offshore installations are electrically well-bonded structures, which normally carry the relatively small [500 mA] leakage currents associated with heavy current equipment. There are tales of significant voltages appearing between deck plates but these are outside the author’s experience. Significant circulating currents have been experienced in the immediate vicinity of the electrical generators on some less well constructed installations but these are very localised and do not affect the major part of the installation.


In normal operation the circuit when connected as illustrated follows the usual rules for intrinsically safe circuits with the screen being bonded at one point only. In this case, multiple bonding of the screen would not cause an incendive current to flow in the screen because the voltage between the bonding points A and B would not be high [50 mV]. It remains good practice to bond the screen at one point in order to avoid low frequency interference being induced in the measuring circuit. In this type of circuit the measuring circuit, the screen and the structure are maintained at or about the same potential by SPD 2 and its bond to point B. The measuring circuit within the field device is usually isolated from the enclosure, [except for RF interference capacitors which are normally negligible from an IS viewpoint], however the SPD is usually bonded to the enclosure and its bond is common with that of the enclosure.


When a lightning strike occurs then a significant current flows in the rig structure, and this current has a fast rise time. [100 kA/ 10ms]. The return path for this current is either the discharge of a locally accumulated induced charge between cloud and sea or part of a circulating current through the earth and troposphere or a combination of both. The usual simplification is that it disappears into the sea or seabed and the rig makes a good connection for it to disappear through. The current sub-divides in the structure and the lower current [10kA] generates a voltage difference in the structure, which is almost entirely due to the inductance of the structure [0,1mH/m] and the rise time of the current [10ms]. Using the suggested figures and assuming a distance of a 100m between A and B gives a voltage difference of 10kV between the two points. The SPDs prevent this voltage difference damaging the measuring equipment but allow a further subdivided current to flow through the interconnecting cable [100A]. It follows that during the short period of this transient current [100ms] the cable is not intrinsically safe, but this is considered an acceptable risk in Zones 1 and 2. An advantage of a well-bonded structure is that the SPD bonds can be kept short [1m] and straight thus minimising their inductance [1mH] which results in a small transient voltage drop [10v] which can largely be ignored. The bonds need to be mechanically robust and secure. The forces generated by these transient currents are not negligible. In practice any transient voltage difference between the field cable and the computer is absorbed by the isolation in the interface.

In normal operation the screen is bonded to the structure and provides an effective electrostatic screen.When the screen is connected to the cable via SPD2 as illustrated, it tracks the potential of that end of the cable during the transient. Consequently there is a potential difference between the screen and the cable at the field end. This is not quite the full voltage difference between points A and B [10 Kv] because the transient voltage is attenuated by the distributed capacitive and inductive of the screen and cable but it is still significant.[5kV] The cable insulation will usually easily withstand this transient but it is important that flashover does not occur at the field termination. It is desirable therefore that the screen should be carefully cut back and insulated and preferably secured in an unused terminal so as to minimise this probability.

The system illustrated describes an effective bonding technique, which is adequately safe in normal and abnormal conditions. A practical installation will almost certainly differ in detail because of the difficulty of ensuring adequate bonding connections. However providing that the basic principle of minimising potential difference is considered then an adequately safe installation should be created.

A non-progress report

June 28, 2007

My apologies for the long interval between entries on this blog, but my contemplating time has been taken up preparing papers for a seminar in Australia, which I rashly promised to provide. If you are involved in Intrinsic Safety now is a good time to get involved in rewriting the standards, which you [or is it possibly just me ?] are always complaining about.

The IEC 31G committee has drafts for comment [CDs] on both the apparatus standard IEC 60079-11 and the system standard IEC 60079-25 in progress. These should be available from your national committee [usually contactable via a trade association] and it may be in your interest to make your views known. Comments made at this stage are the most effective because it is easier to make changes and introduce new concepts.. Later the document acquires its own ‘momentum’ and changes are less acceptable and tend to be deferred until the next edition. This introduces a five to ten year delay, which is not ideal but almost inevitable. A system of maintenance teams was introduced to speed up the process but the interval between editions is still about five years.

Any comments will be discussed in Kuala Lumpur in November this year. There is also a voting document [CDV] on the Fieldbus standard IEC 60079-27 but while this is passed the point of fundamental changes, it is worthwhile reading if you are involved

There is some thought being given to the subject of the risk presented by combined gas and dust hazards. There are a number of questions were the required technical knowledge is not available and consequently anyone who can make a contribution in this area would be welcomed with open arms (Contact your local committee)

The whole process of creating IEC standards is far from ideal, but is the only system in existence and the known alternatives are no better or even worse. Inevitably the majority of the delegates to IEC committees have one or more of the following characteristics. They are passed their sell by dates, have particular interests to defend, have large egos and epitomise their national stereotypes. The selection process is not ideal. What is needed is some new blood with new ideas to counteract the existing inertia and combine with the more experienced participants to produce better standards. The negative side of participation is that it takes time and attending meetings is expensive. The positive side is that some of the participants are worth knowing and are good company, plus the meetings tend to be in interesting places !