Yes, this pipeline is much more dangerous. Here’s exactly why.

I am surprised that I still encounter people who are indifferent about the Dragonpipe (Mariner East  pipeline system) because “we already have lots of pipelines around here, and there’s never been a problem”. Thanks to the “Canary” software that was leased in conjunction with the recent Citizens’ Risk Assessment, it is now possible to show exactly how much more dangerous the Dragonpipe is.

For those who are not familiar with Canary, it is the software that was used by Quest Consultants (the vendor that performed the risk assessment) to model the consequences of a leak. The inputs to the software are factors such as pipeline diameter, size of leak, contents of the pipeline, interval until the gas is ignited, wind speed, and so on. When the software is run, it provides numeric data about the consequences of the resulting fire, as well as diagrams showing how extensive it is. Some of those diagrams appear below, comparing the consequences of a leak involving “natural gas liquids” (NGLs) with a leak involving ordinary natural gas.

Stages of a pipeline leak and explosion.  Before we get to the diagrams, though, it is important to bear in mind the three stages through which a leak progresses when there is a pipeline fire.

Stage 1: The flammable vapor cloud. If the gas doesn’t immediately ignite when the leak or rupture occurs, the gas forms a vapor cloud that expands rapidly.  After a few minutes (often around 5 minutes in the variations I have tested), it reaches its maximum size.

Why doesn’t it continue to expand beyond that maximum? Although gas continues to spew from the pipeline for hours, the actual area that is flammable does not grow. That’s because, at the edges, it mixes with the surrounding air and becomes too diluted to burn. So an equilibrium is formed: gas is constantly added to the cloud from the leak, and it is constantly diluted at the edges of the cloud. The boundary where the flammable gas becomes too diluted to burn is called the “lower flammable limit” (LFL). The LFL defines the outline of the flammable vapor cloud.

Stage 2:  The flash fire. When the flammable vapor cloud encounters an ignition source, the entire cloud burns within a few seconds. This is the “flash fire”, and it is the deadliest and most destructive phase. Anyone outdoors within the cloud will not survive, and flammable structures will be set on fire.

Stage 3: The jet fire. Once the vapor cloud has burned, the fire continues as a “jet fire”, burning the gas as it emerges from the pipeline in a manner similar to the operation of a blowtorch. The heat is intense and sustained, but the affected area is far smaller than the area of the flash fire. The jet fire can be a problem for emergency responders because it can’t be extinguished until the contents of the pipe have burned off. And, depending on exactly where it is, it may block access to the area that burned in the flash fire.

If ignition occurs immediately when a pipeline begins to leak, steps 1 and 2 are skipped. There is no vapor cloud and no flash fire, only a jet fire. That explains why the risk of fatalities is greater when ignition is delayed.

The size and shape of the vapor cloud: comparing natural gas and NGLs. Given that the flash fire presents the biggest risk, it is important to know the size and shape that the vapor cloud can take. This is where the Canary software comes in. It calculates an engineering model of the flow of the escaping gas and how it mixes with the air. The model has been tested with field experiments to confirm that actual vapor clouds behave as predicted.

I have done modeling runs with Canary to see what happens when a pipeline containing NGLs ruptures, compared with the same rupture for a pipeline containing natural gas.

Three features of the gases are important in determining the risk from a leak or rupture.

1. The fact that natural gas is lighter than air, and it rises; while NGLs are heavier than air, and they spread along the ground,
2. The fact that NGLs are under much higher pressure in the pipeline, and
3. The flammability of the gases themselves.

In my modeling I have been struck by how important the first of these three features is. Because natural gas rises, most of the vapor cloud it creates is high in the air. In rising, it also mixes more readily with the air, so the vapor cloud is much smaller. The flash fire resulting from a higher, smaller vapor cloud is far less dangerous to people on the ground nearby.

You can think of escaping natural gas as behaving like steam from a tea kettle: even if it spurts out horizontally, it will bend upward. NGLs, on the other hand, behave like fog. They spread out and settle in low-lying areas.

Vapor cloud shapes for a 12-inch pipeline. This difference in the behavior of the gases is very obvious in the output from the Canary software. When I modeled the 12-inch pipeline in the vicinity of Shamona Creek School, I decided to do an experiment. I modeled a release of natural gas and compared it with a release of propane (one of the NGLs). In both cases, I assumed a total rupture of a 12-inch pipeline. I also assumed a relatively calm day (2 mph wind speed), which makes the cloud disperse more slowly.

The comparison shown here for the 12-inch pipeline would be similar if I had modeled a 16-inch or 20-inch pipeline: in each case the risk posed by NGLs is dramatically greater than for natural gas. The area involved would be greater, the larger the pipeline, of course.

Here are the 12-inch results, in diagrams generated by the Canary software. The leak is located at the left edge of the diagram in all cases, and the downwind direction is to the right.

First, here’s an overhead view, showing how propane spreads out but methane does not.

 

Looking at the two clouds from the side shows how methane rises, while propane spreads out across the ground.

These differences in the behavior of the gases mean that the risk to people on the ground is very different in the two cases. When the methane vapor cloud ignites, only people very close to the pipeline—within 100 feet or so—are endangered, and only if they are directly downwind. By contrast, ignition of the propane vapor cloud could kill people almost 1500 feet from the pipeline, and they need not be directly downwind. (The comparable distance would be over 2100 feet in the case of a 20-inch pipeline.)

Stated in terms of acreage, the flammable methane cloud shown here covers less than a tenth of an acre (and a good bit of that is high in the air), whereas the propane cloud covers about 30 acres, all at ground level.

Our first responders have emergency plans designed to protect people within a few hundred feet of a methane release. How will those plans work in trying to cope with a 30-acre flammable fog whose boundaries shift with the wind?

Conclusion: the difference is immense. So the next time someone tells you this pipeline isn’t that different from the natural gas pipelines we already have, refer them to this blog post.

Are they ready to accept that a flammable cloud covering a tiny fraction of an acre and one covering 30 acres represent similar risks? Which would they accept in their neighborhood or near their child’s school?

Update, 12-7-18: The Delaware County risk assessment. which was released after this blog post, reinforces these conclusions. It looked at both the 20-inch Dragonpipe and the 18-inch Adelphia natural gas pipeline. In the case of a rupture of the Dragonpipe, it projected a vapor cloud and flash fire that would actually be about 8 times larger than that indicated by the Citizens’ Risk Assessement (on the order of 250 acres). For the Adelphia pipeline, by contrast, the potentially fatal area would be less than 2 acres (with most fatalities from the heat of the jet fire, not the flash fire).

Again, the results indicate that the consequences of a rupture of an NGL pipeline are something close to 100 times worse than the rupture of a natural gas pipeline. These are completely different categories of pipelines. Don’t let anyone tell you they are similar!