Combustible Dust: Probability of Ignition

“Before the first atomic bomb test, scientists took the time to calculate whether the blast would ignite the nitrogen in the Earth’s atmosphere and incinerate us all. The risk was low and the test went off, but Rees wonders what the odds would have had to be to discourage the bomb makers.”  — Dennis Overbye

Many like to quote the old adage, “Given a fuel and air, ignition is free.” As with all adages, it is pithy, but not to be confused with absolute. Anyone who has ever been camping and tried to start a campfire knows this. Moreover, refineries experience spills quite frequently that don’t burst into flames.

This is why we use probability of ignition.

Probability vs. Frequency

Normally, we talk about the probability of ignition rather than the frequency of ignition. That is, what is the probability, given a combustible mixture of fuel and oxidizer, that there will be an ignition of that mixture. We apply the frequency to the presence of a combustible mixture. Frequency of ignition would make sense, however, if the combustible mixture is always present.

When a condition is continuously present, that is, in the high demand mode, most practitioners use the advice in Appendix F of Layer of Protection Analysis: Simplified Process Risk Assessment, and consider the frequency of the event to be the probability of the event times once per year. So, a probability of 0.1 in a low demand mode would be expressed as a frequency of 0.1/year in a high demand mode.

Probability of Ignition in LOPA

One of the most common places for using probability of ignition is as an enabling condition in Layer of Protection Analysis (LOPA). In 2014, the Center for Chemical Process Safety (CCPS) published a book, Guidelines for Determining the Probability of Ignition of a Released Flammable Mass. It goes into great detail about the mechanism for ignition, the factors that can be considered in an estimation, and sources of data. But LOPA is supposed to be a “simplified process risk assessment.”

When Bluefield Process Safety was developing its LOPA tool, it included several causes, frequency modifiers, and IPLs with their associated values.  The values Bluefield used for the probability of ignition included:

0.1   Pool or jet fire ignited by low-energy source

0.3   Pool or jet fire caused by or near a high-energy source

and

0.1   Small vapor cloud ignited by low-energy source

0.3   Small vapor cloud ignited by high-energy source

0.5   Large on-site release of vapor cloud

0.9   Large release of vapor cloud to area with uncontrolled ignition sources

Implicitly, Bluefield considered three factors: immediate ignition of a liquid or gas vs. delayed ignition so that a cloud formed, high-energy ignition sources vs. low-energy ignition sources, and in the case of clouds, the size of the release.

Bluefield based the values that it used on a very conservative reading of section 16.10.3 (pg. 1213) of Lees’ Loss Prevention in the Process Industries. Lees’ refers to the Cox methodology, which gives these as typical ignition probabilities:

Gas releases

• Minor release (≤ 1 kg/s): 0.01 (or 1%)
• Major release (1–50 kg/s): 0.07 (or 7%)
• Massive release (≥ 50 kg/s): 0.30 (or 30%)

Liquid releases

• Minor release (≤ 1 kg/s): 0.01 (or 1%)
• Major release (1–50 kg/s): 0.03 (or 3%)
• Massive release (≥ 50 kg/s): 0.08 (or 8%)

Interestingly, both Lees and Cox concerned themselves with flammable vapors and gases and flammable liquids. They did not address combustible dusts.

Probability of Ignition for Combustible Dusts

So, how does the probability of dust ignition compare to the probability of vapor, gas, or liquid ignition? This is where the minimum ignition energy (MIE) comes into play.  The higher the MIE, the less likely that a potential ignition source will be able to ignite the fuel.

Let’s compare the MIE of flammable gases, flammable liquids, and combustible dusts.

The MIE of flammable gases is typically around 0.25 mJ, although there are notable exceptions. Carbon disulfide, for instance, is 0.015 mJ, and acetylene and hydrogen are both 0.017 mJ.  At the other end of the spectrum, ammonia is 680 mJ.

The MIE of flammable liquids is similar to that of flammable gases, or a little bit higher. Gasoline, for instance, has an MIE of 0.8 mJ. Flammable gases and flammable liquids are similar enough (with the notable exceptions) that is reasonable to treat them the same in a LOPA.

The MIE of combustible dusts is considerably higher, ranging from 1 to over 10,000 mJ.  (Sulfur dust, with an MIE of 0.3 to 3 mJ, is a notable exception.)  Hence, it is easy to justify decreasing the probability of ignition for combustible dusts by at least an order of magnitude, and depending on the dust, two orders of magnitude.

That means that if an ignition probability of 0.1 is conservative for flammable gases and liquids with an MIE of 0.3 mJ, then 0.01 would be conservative for combustible dusts with an MIE in the range of 1 to 10 mJ. An ignition probability of 0.001 would be conservative for combustible dusts with an MIE > 100 mJ.

This All Assumes That Ignition Sources Remain Controlled

It remains important to control ignition sources.

Ignition sources include low-energy sources, like static electricity, mechanically generated sparks, and some electrical equipment, and high-energy sources, like hot surfaces and open flames.  The static sparks we feel in the wintertime are around 0.5 to 1.5 mJ. The energy we feel from an electroshock weapon, i.e., a Taser, is over 10 mJ. A corona discharge has an ignition energy of about 70 mJ.

It is the ignitable nature of electrical equipment that has led to the requirement for electrically classified equipment—Class I for flammable gases and vapors, Class II for combustible dusts, and Class III for ignitable fibers.

As for high-energy ignition sources, any surface hotter than the autoignition temperature of the flammable or combustible material can ignite it. Open flames, with temperatures of 600 C for small wood fires up to 3,000 C for a properly tuned oxy-acetylene torch, will easily exceed the autoignition temperature of most materials, and so are high-energy ignition sources.

The probabilities of ignition mentioned here all assume that measures are in place to control ignition sources. This means grounding and bonding for equipment, classified electrical equipment suited to the hazard location classification where it is used, and hot work permits to control the presence of open flames and hot surfaces.

But even providing all of this is not a guarantee that ignition cannot occur.  Everything fails. That is why we talk about the probability of ignition.  How low that probability must be depends on each organization’s risk tolerance criteria.