Marcellus (PA) - update until 2016-04

This is the first presentation about the Marcellus, in Pennsylvania. Also, it is the first time I present natural gas production, as almost all hydrocarbon production in Pennsylvania is gas. And lots of it; according to the EIA it ranks 2nd, only behind Texas, and it is currently responsible for roughly 20% of total US natural gas production. Together with the Utica play, the Marcellus has provided 85% of the growth in natural gas production in the US, since 2012.

In the presentation, the production from all the unconventional, horizontal wells is shown, as reported by the PA DEP. In total almost 6700 horizontal wells have started since 2010, until April this year. As you can see in the “Well Status” tab, there also appears to be a large number of DUCs. These wells have all been spudded, but no reported production yet.

Natural gas production has grown significantly, and reached a new high in February this year, just over 14k MMcf/d (=14 Bcf/d).

You can see in the “Well quality” tab that average well productivity has been steadily rising over the years, but seems to have halted since 2014. As I would expect, it appears that natural gas production declines are a little lower compared with the declines in LTO production that we’ve seen in the other basins.

I have adjusted the presentation, such that it’s easy now to see both gas, and oil (& condensate) production. In each slide, there is a selection (“Product”) that allows you to switch between the two.

Although the Pennsylvania state data was pretty good, there are a number of limitations:

  • I could not find in which formation each well is located, and therefore currently assume that they are all in the Marcellus. This is incorrect, as some of them are in the Utica and Point Pleasant formations. If someone has an idea how I could distinguish between them, please let me know.
  • Prior to 2015, production of wells is not reported monthly, but (semi-) annually. I used an algorithm to assign this (semi-) annual well production to individual months, taking into account the likely date that a well started production, and the typical production profile.

As this is the first time I report Pennsylvania production, and natural gas production, the chance of issues is somewhat higher. If you notice something strange, or if you know this area well and the data presented confirms with your expectation, I would like to hear it.

On Tuesday (edit: Wednesday) I plan a next presentation, on a projection of production from existing LTO wells over the coming 2 decades. Next week Friday I expect a new update on North Dakota, after which I will be on a 2 week holiday.


For this presentation, I used data gathered from the following sources:

  • PA DEP


The above presentation has many interactive features:

  • You can click through the blocks on the top to see the slides.
  • Each slide has filters that can be set, e.g. to select individual or groups of operators. You can first click “all” to deselect all items. You have to click the “apply” button at the bottom to enforce the changes. After that, click anywhere on the presentation.
  • You can switch between different products (e.g. natural gas, or oil), by using the “Product” selection, present in all tabs.
  • Tooltips are shown by just hovering the mouse over parts of the presentation.
  • You can move the map around, and zoom in/out.
  • By clicking on the legend you can highlight selected items, and include or exclude categories.
  • Note that filters & selections have to be set for each tab separately.
  • The operator who currently owns the well is designated by “operator (current)”. The operator who operated a well in a past month is designated by “operator (actual)”. This distinction is useful when the ownership of a well changed over time.
  • If you have any questions on how to use the interactivity, or how to analyze specific questions, please don’t hesitate to ask.


  • Nony says:

    1. Muchos gracias, you stud! You figured out another data source.

    2. And thanks for looking at gas, even though you love oil more. I do think there are insights to learn from looking at each in compare/contrast. Even though different.

    3. (Not a criticism, just nailing down a few caveats.) Given this is PA only, significant gas in WV will be not counted. (Also some in Ohio, little from the Marcellus and lot from Utica). It’s a little bit like situation of NM in Permian or MT in Bakken.

    4. Yeah, agreed on the formation caveat. I actually think of this area more and more like the Permian in terms of stacked production. You have pads in SW PA with both Utica and Marcellus wells. (Sometimes Upper Devonian on same pad.) So they share infrastructure. They are using common pipes also, although eventually it will be desirable to have a high pressure Utica system side by side the low pressure Marcellus one.

    5. Given 3 and 4, I think of the whole region more and more as “the App” (Appalachian) since what matters for pipelines and prices is really total production regardless of Utica vs Marcellus (like Spraberry vs Wolfcamp). The dimension of NE PA versus SW PA is probably more meaningful than WV versus SW PA from an economics standpoint (maybe not data gathering standpoint) since the two parts of PA are somewhat isolated from each other in addition to the whole thing have limited connection to Henry Hub.

    6. The slow(er) decline is fascinating. Terry Engelder argued several years ago that this would happen based on data from old verticals from the 70s-80s gas shale project wells. Looks like he was right. (Opposite was an expert that I won’t name per your rules, but could link to, who held up the Haynesville declines as indicative of the Marcellus future.)

    7. Look at Cabot 2014 wells, cum, late in time. Funky rise upward. Is that an artifact? Or shutins?

    8. I prefer just “oil” versus oil and condensate (condensate is oil). Or you could say crude and lease condensate to be hyper precise. (Plant condensate is another beastie and won’t show up in well production numbers as it is the “second squeeze” out of the gas stream.)

    9. Mighty, mighty, MIGHTY! Go Marcellus shale!

    10. If you have not watched this before, might be interested in it. Just guy seems very logical about how he tries to attack analysis.

    1. Enno says:


      Thanks for the comment.
      Coming back to some of your points:

      3. Yes, gas production in PA is most significant, but Ohio and WV are also interesting. Unfortunately, as Gerard mentions later, WV only reports annually. I may do a seperate post on WV once, and may look at Ohio earlier, also because LTO production is quite high there. Interestingly, glancing at the data made available in Ohio, it looks like they are dumping the full database online, incl 10s of tables and database views that are normally not shared. Will be a little work to figure it out, but at least it seems like all data is available (although apparently no monthly production numbers, so I have to employ a similar algorithm as used for PA).

      5. I agree. That’s also a reason I group NM Permian, and TX Permian. There is only a technical reason behind this, which is that each state reports very differently.

      6. I found it striking how the average declines seem to follow an exponential function (easily seen if you switch to “log” mode).

      7. All data since 2015-01 is reported, without any intervention, so I suspect that operational reasons are behind this.

      8. Good point, I will think about this.

      10. Thanks for the link, I will watch it once I got a little more time.

  • John says:

    Thank you Enno. This is very appreciated by me and surely by many others.

    I would love to see Art Berman’s regional map with the wells shown by operator, year of completions D Cummulative production.

    I am a big fan of both you and Art. Thank you again.

    1. Enno says:

      I appreciate it John, thanks.

    2. Gerard McHale says:


      Great stuff, as usual. I applaud your ongoing efforts as you continue to expand your horizons.

      The site contains extensive amounts of data on each of the 15,000+ permitted unconventional wells in Pennsylvsnia going back to 2008 or so. Info includes targeted formation (20 or so recognized by DEP), as well as days online – helpful in the twice-yearly reporting timeframe.
      The Utica production in PA is, at the moment, somewhat inconsequential, as there are only a dozen wells in Tioga county operated by Shell (SWEPI), and the dozen, attention grabbing Deep Utica wells in western Greene and Washington counties. Westmorland county also has one, the Gaut from Consol.
      Here again, the recent single Utica well in Potter county, JKLM’s Sweden Valley, may prove highly significant as it continues to expand the Utica’s productive footprint.

      I’ll touch base tonight with some info related to Ohio, West Virginia, and the shallower Upper Devonian formations which currently impact the PA production data.
      Good luck with your ongoing efforts compiling this data. It’s an impressive bit of work.

      1. Enno says:


        Thanks for the compliment and info.
        The PA DEP already reports online days, which I indeed also use to get to more realistic monthly numbers.

        I will check out the site you mentioned before my next update on Marcellus.


  • alex says:

    Thanks Enno,
    great data as always.

    It is interesting to see how the ‘time decay’ of production gains is reducing each year. i.e. if we take production at the end of 2010 and add 2011, it then took ~4 years for the cumulative total incl. 2011 to equal end 2010 production again. So 2011 added 4 years worth of increased production.

    – End 2010 production = Cumulative 2011 (~4 years later)
    – End 2011 production = Cumulative 2012 (~2.5 years later)
    – End 2012 production = Cumulative 2013 (~2 years later)
    – End 2013 production = Cumulative 2014 (~1.5 years later)
    – End 2014 production = Cumulative 2015 (~1.4 years later)

    Given the large capex cuts at many of the major Marcellus producers in the last 12 months, it won’t be surprising to see 2016 cumulative below the end of 2015 level within less than 12 months given the above trend. i.e. Marcellus output will drop in 2016.

    Across all vintages production dropped quicker then usual in mid 2015, and then rose notably into the winter. Can anyone explain that one? Given it affected every year, this wasn’t just because of new wells coming online. And I haven’t read anything to suggest there was wide spread re-fracking/enhanced recovery going on. It’s more like all producers choked back mid 15 (hoping for higher pricing?), and then opened the valves again for the winter. The reason I think this could be interesting, is that I have read a lot of commentary talking about record production gains in the Marcellus in early 2016. But if that winter 2015/16 bump was across all well vintages because of choking, then the bump was just displaced output, and 2016 production might surprise on the weak side as the start of the year wasn’t as strong as many thought.

    1. Enno says:

      Thanks Alex,

      interesting observations.

      Production indeed shows some unnatural fluctuations in 2015/2016. I don’t know the reasons behind this; but these are indeed in the raw data, as data for 2015 and onward are provided on monthly basis, unlike earlier data.

    2. Nony says:

      alex: good analysis. Yes, they did shutins (for price), to open up in winter. This has been widely reported. Also, it is actually not that strange a practice–this is how natural gas was managed in the 70s-80s, rather than all the use of storage. IOW, the field itself is a form of storage. [One more nuance” to understand volume-price relations in the Marcellus, it is critical to use local App pricing, not Henry Hub. The pipeline constraints mean that the markets are disconnected and you can’t even assume steady basis difference.]

      1. Joseph Wells says:

        Nony, makes sense. Marcellus has a long lag between drilling and production. Don’t know the breakout between choking, shutting-in, and being shut-out by pipeline constraints. My modeling suggests an average of 52 weeks between spud and production compared to an average of about 18 weeks at all other shale basins. In the attached I calculated production at Marcellus assuming both a 24 week and a 52 week lag from spud to production.

        I’m using DPR and Baker Hughes data. For future rig counts I just assumed that Marcellus rigs ramp up to 60 over the next 12 months and then hold constant. Production is dropping through early 2017.


  • Coffeeguyzz says:


    Just checked the Ohio DNR site and they only show 29 Marcellus producers in the state as of last week.

    While this undertaking of the PA Marcellus is certainly a lot of work, along with the other formations you are compiling data on, the Appalachian Basin, as Nony has mentioned, may come in time to be viewed somewhat as a whole, although there are certainly different characteristics, and, indeed, are distinct formations entirely.

    The Ohio site – “” is by far the cleanest site I’ve come across to get data.
    Clicking on the “Shale Activity” button on home page (little grey button to the right) brings you to the downloadable buttons where quarterly production is presented in a single sheet, Excel format, alphabetized by producer.
    The biggest discerning/interpretive tack may arise from recognizing the oil/condensate/dry gas wells as they have widely differring characteristics.

    Before posting this comment, I spent about 2 minutes to learn the 1,352 producing wells in Qtr 1, 2016 had precisely 50 that put out over 1 Bcf during the quarter, a staggeringly high amount indicating the richness of this area.

    The West Virginia folks still post annualized reports. The 2015 numbers were released three weeks ago in Zip compression which I still cannot open.
    Virtually all WV production stems from the Marcellus, although it has a LOT of NGL content.

    Although operator presentations are regularly viewed with skepticism, if not outright disdain, by many, their presentations are a great source to get a ‘sense’ of what this area is all about in the operational perspective.
    Companies like Rice, Antero, Range, EQT, and Consol provide a wealth of technical, on the ground data with which one can develop a sense of familiarity.

    Once again, thanks for your contributions.
    Will look forward to the next one.

    1. Enno says:

      Thanks Gerard,

      Indeed Ohio shares a remarkable set of data, although unfortunately it lacks some precision (no monthly production data, but quarterly/annually). I may include it at some time in the future.

      I saw WV indeed released 2015 data just a while ago. I have no problems with the zip file.

      “The biggest discerning/interpretive tack may arise from recognizing the oil/condensate/dry gas wells as they have widely differring characteristics.”

      Formerly I focused on oil wells. But as I report gas production as well now, I will no longer filter horizontal wells based on their “type”. It is easy to for the viewer to switch between gas & oil production (in the “Well quality” tab), so that the total production from the selected wells can be seen in order to judge their productivity.


      1. Coffeeguyzz says:


        Quick correction my Utica numbers for PA.
        Seems there are actually about 50 Utica wells in the northwestern counties (Lawrence, Beaver, Butler, Mercer) but the output seems under 150 MMcfd.
        With the exception of Hilcorp and Rex, no one is too active up there.

        The Upper Devonian formations, according to a study by Wrightstone Energy last year, has about 75 PA wells producing, with a relatively minor statewide contribution.

        To precisely identify the source formations is probably WAY too time consuming considering the small relative output, but that is apt to change in years to come as the Deep Utica is developed.

  • Enno says:

    The US shale projections post I planned for today needs a little more work. I aim to finish it in the coming 2 days.

Leave a Reply

Your email address will not be published. Required fields are marked *