On Friday, October 19, the full text of the Citizens’ Risk Assessment was published. You can download it here. This report analyzes the risks of the Mariner East 2 pipeline project and reports them as actual risk numbers: the probability of fatal accidents per year. It represents the culmination of an effort initiated by State Senator Andy Dinniman and carried out by numerous volunteers who developed the request for proposals, vetted and selected the vendor, raised the necessary funds, and made sure the results were broadly circulated.

This report is the product of a remarkable undertaking by private citizens, made necessary by Sunoco’s policy of secrecy and the failure of local municipalities and state regulatory agencies to take this project on. (Delaware County is an exception: it has commissioned its own risk assessment, due out in November. Middletown Township has also retained a pipeline hazards expert to do a local hazard assessment—but he will not attempt to assign probabilities as a “risk assessment” does.)

In this post, I will briefly review the results of the study and then look in more detail at the potential consequences at one particular site (near the valve station at Glenwood Elementary School in Delaware County).

Finally, I will take a look at what the software used in the risk assessment can tell us about a possible worst-case accident at the Glenwood location. This software, called Canary, is currently available on lease, but fundraising is needed to get a permanent license. When you read about the Glenwood analysis, I think you’ll see why I think it is so important to purchase it.

Highlights of the risk assessment report. The final report goes into much more detail than the summary slides that Jeff Marx of Quest Consultants presented in August, but the key take-aways are the same:

  • Heightened risk exists in the vicinity of valve sites.
  • Heightened risk exists near HDD entry and exit points.
  • Two pipelines produce approximately double the risk of a single pipeline. Three pipelines approximately triple the risk, and so on.
  • These pipelines (based on industry-wide failures of pipelines carrying highly volatile liquids, and not considering Sunoco’s industry-worst leak record) are likely to average a leak every 2 or 3 years, statewide.
  • Even the smallest possible leak in an NGL transmission pipeline can result in fatal fire or explosion.
  • For the three specific locations studied in detail, the study shows exactly which homes and buildings are in harm’s way, and how serious their risks are.
  • It shows that, for those in the immediate vicinity of these pipelines, death from a pipeline accident is about 10% as likely as death from a car accident, and about 150 times more likely than death from a lightning strike.

Study limitations. There are many important lessons to be learned from this study, and I don’t want to minimize that fact. But the study has a number of limitations as well, and you need to know about some of those.

  • Individual risk vs. population risk. The report does not distinguish between risk to a single person and risk to a group (e.g. a school or a mall).
  • “Black swan” events. Some events are so rare that there are no good statistics about them. These are ignored in the report, but we still need to consider them.
  • Death risk only. An accident would cause many injuries, including life-changing burns, but these are not studied.
  • Flat terrain assumed. The software does not take pooling in low-lying areas into account.
  • One pipeline accident might damage a neighboring pipeline. This possibility was not studied.
  • Other risks not studied: calm winds (which could result in bigger explosions), toxic smoke from fires triggered by an explosion, asphyxiation hazard.

If you are interested in learning more about these limitations, you can review them (and some of the reasons for them) here. The important thing to know is that these additional factors make the risk greater than the numbers reported in study. How much greater? That’s unknown.

What does this mean for the Glenwood neighborhood? The conclusions reached by the risk analysis are all important facts to know, but they are a bit abstract. To provide some context, let’s consider what they mean for one of the sites that was studied in detail, the area near Glenwood Elementary School in Delaware County.

The image below, taken from the report, shows the area near the Glenwood school.

glenwood risk (from QRA) 10-21-18
The area near Glenwood Elementary school. The thin magenta line represents the route of Mariner East 2. It is a dotted line in areas where HDD is used. The orange circle represents the highest-risk area (a valve site). Magenta, blue, turquoise, and green lines represent successively lower levels of risk. See text for details.

In this image, the various levels of risk are indicated by the colored contour lines. The small orange circle represent the area of highest risk (about one chance in 10,000 per year of a fatal accident). At the center of that circle is the above-ground valve site behind the Glenwood playground. This level of risk (1 in 10,000/year) is generally considered unsafe by international standards. The US has no similar safety standards.

As a reminder: in this study, the “risk of fatality” is the risk of any death at a location. It does not distinguish between the tragedy of a single death at that spot and the catastrophe of the death of dozens of children who happened to be at that location at the wrong time.

The magenta ovals represent the next-highest risk areas (about one chance in 100,000 per year of a fatal accident). One of the ovals surrounds the valve site. The other two are at entry/exit points where horizontal directional drilling (HDD) is used to put the pipe deep underground. These areas have elevated risk because gas leaking anywhere on the HDD line is likely to follow the pipe to the surface. On the other hand, above HDD areas (the dotted magenta line) risk is reduced because underground gas leaks there will probably follow the pipe to the surface and will not surface directly above the leak.

It should be noted that Sunoco’s attempt to drill from Riddlewood (near Glenwood School) south to the Tunbridge Apartments failed, and the company has abandoned that site, at least for the time being. If it is permanently abandoned, the magenta oval at the lower right would be eliminated. At this point, we have no idea what Sunoco’s plans for that part of the pipeline may be.

As you can see from the diagram, the risk to the school is considerably lower than the risk right at the valve site, but the school is far from being completely out of harm’s way.

Although it is not obvious from the image, tree cover affects the risk—and not in the way you might expect. It turns out that the turbulence created when gas moves through a forest results in a flame front that moves faster and burns more violently than gas in an open area. The explosive force released is greater. So, in areas where there is tree cover in the vicinity of the pipeline, the risk of fatality is increased.

Sunoco’s changing plans. Since this study was completed, Sunoco’s plans have been shifting. It has run into problems with HDD at several sites and is therefore planning to use an old 12-inch pipeline that previously carried gasoline and heating oil to bypass the parts of ME2 and ME2x that are not complete. This possibility was not considered in the risk assessment.

If this 12-inch pipeline is put into service, it will represent an immediate danger for the Glenwood area. Unlike ME1 (which is already in operation and which runs to the west of Glenwood), the 12-inch pipeline uses the same easement and valve site as ME2, located right behind the school. The map below shows this situation.

12-inch route in Delco7-7-18
If Sunoco’s plan for the 12” line is implemented, it would carry NGLs from northern Chester County, splitting off from the ME1 right-of-way at Forge Road, and continuing southeast to the to the valve station behind Glenwood School. There, it would connect to the Dragonpipe, which would carry the NGLs to Marcus Hook. This would bypass the incomplete parts of the Dragonpipe in Chester and Delaware counties.

The Canary software and implications of the 12-inch pipeline. In connection with the risk assessment, Del-Chesco United got the opportunity to lease the software package (called “Canary”) that Quest Consultants used to estimate the consequences of leaks on the Mariner system. This software can be used to calculate the consequences of any leak on any pipeline.

I was among those who got training on the Canary package, and I recently used it to model the effect of a rupture of the 12-inch pipeline near the Glenwood valve site.

The user of the software enters values for a long list of variables, including the size of the pipe, the size of the leak, the material being transported, its pressure, the wind speed, and how soon after release the gas finds a source of ignition.

Then, the software calculates the consequences, based on an engineering model of pipeline dynamics and burning gases. For our purposes, the main consequences are a cloud of flammable gas (whose size increases rapidly), a “flash fire” (in which the entire cloud is burned in seconds, once it finds an ignition source, and finally a “jet fire” (a somewhat smaller but hotter blowtorch-like fire that emanates from the broken pipeline and continues until the gas in the pipeline is used up, which would take many hours). There is more detail about this in my recent case study of a possible rupture near the Wellington senior living center in Chester County.

I applied the Canary software to a worst-case rupture of the 12-inch line near the Glenwood valve site, with the results shown in the image below.

glenwood worst-case 10-21-18
The light-colored area in this image represents the area that would be enveloped in a “flash fire” in a worst-case rupture of the proposed 12-inch pipeline. See text for details.

I assumed a 2-mph breeze blowing from the rupture site toward the school, and I assumed it would take 10 minutes for the gas to find an ignition source. The light-colored area shows the extent of the gas cloud and the ensuing flash fire. It would stretch 1500 feet from the pipeline rupture. The width of the cloud would be 1640 feet. The entire school would be vulnerable to this fire. Anyone outdoors would be killed instantly. Being indoors would provide some protection, depending on how much gas had penetrated the building. The flammable parts of the school building, as well as surrounding homes and businesses, would be set on fire.

If the gas did not ignite within the first few minutes, it could flow downhill, along Lenni Road or down to Chester Creek. The gas would be concentrated in the valleys, presumably presenting an even greater threat there. This is a serious worry, but the Canary software can’t easily handle that situation, so I didn’t model it and I won’t discuss it further.

At the top left of the image, you can see Granite Farms Estates, a senior living facility. As with Glenwood School, the entire facility could be engulfed in a flash fire in the event of a pipeline rupture. Depending on where the rupture occurred, the resulting jet fire could block the only entrance to the facility for many hours, preventing emergency personnel from rescuing the residents. I discussed this same access problem in my post about Wellington.

The Glenwood School case that I modeled represents a rare event. The same could be said of a similar event at Granite Farms. But we know they could happen. Similarly rare events have happened in Follansbee, WV, at Brenham, TX, and other locations. We need to have an accurate idea of what the consequences would be if they did happen. This software is a good start.

Help us keep this software. That brings me to my final point: we are currently leasing the Canary software because we don’t have enough money for a permanent license. We’ll need to raise another $7,000 to get one. But we will continue to need this software, for studying both this pipeline and for others that will be proposed for our area. Please help us raise the funds to keep it, by going to this site:


Thank you!