A Guest Post by George Kaplan
Part I Discoveries and Reserves
The following are a few charts and observations concerning the US Gulf of Mexico production, with data mostly taken from BOEM for the OCS and a bit from EIA. The BOEM data site has been updated and makes it easy to get the raw data, but not in a very easily discerned way. This post is intended to make it a bit easier to follow and come to your own conclusions concerning the likely future of GoM production.
I’m sure I’ve made a few mistakes in getting and downloading the data, but the relative trends are probably more informative than absolute values so there will need to be follow up revisions as new data is available and hopefully any issues will get smoothed out. Reserve data is available for 2015. New data for 2016 is likely not available until the end of the year, but when it is it will be interesting to see what changes there are. The main focus here is on oil but gas data is also presented. Production data is issued twice monthly but some, such as for BP, is up to six months late so I’ve just assumed daily flow stays constant to fill in any gaps.
Every year BOEM issues a report for the previous year’s reserves (i.e. December 2016 covers 2015): ”https://www.boem.gov/Reserves-Inventory-Program-Gulf-of-Mexico-OCS-Region/”
The reserves given are 2P, with the description:
“Reserves in this report are proved plus probable (2P) reserves estimates. The reserves must be discovered, recoverable, commercial and remaining. Reserves, starting with the 2011 report, now include Reserves Justified for Development. “
The first figure below shows cumulative backdated oil and gas reserves by discovery year (shown as produced and remaining, and stacked to give total discovery). Finding new oil and especially gas is now a struggle. However one problem with the chart may come from the way BOEM record production against lease but reserves against fields. Lease and fields and fields are associated with exploration blocks, but also split by ownership. One field may cover several leases, and that is easy to accommodate but equally one lease may contain several fields and then, as far as I can tell, but there may be arcane knowledge to which I am not privileged, the reserves are logged against a common field number. Hence later discoveries might be recorded against older fields. For example recent tie-ins for South Deimos and West Boreas to the Olympus Spar are not shown as separate fields but rather against Mars-Ursa, and would show as reserve growth on the older discovery. A bigger impact is that there are discovered fields, mostly since 2010, which do not have confirmed development plans yet, and so do not get counted as reserves by the SPE methods that BOEM uses. Therefore the flattening in new discoveries is not really as severe as shown, at least for oil. However, as discussed later, discoveries currently are declining towards zero.
Condensate is included in the oil numbers and associated gas from oil fields with the gas reserves. NGLs are also included with the gas. The GoM is, or has been, particularly gas prone so these represent significant proportions of the total reserves, so it will only increase as a result of newly sanctioned projects, while there are some available.
Note this chart shows the situation in 2016 – i.e. the backdated reserves discovered and remaining for all the years up to the one shown; it isn’t a record of cumulative production to that year. Total discoveries for oil so far total 23.1 Gb and for gas 34.5 Gb (based on conversion rate of 5600 mcf / bbl). The curves above are really made up of a series of similar shaped asymptotic growth curves for shallow, deep, deeper – Jean Laherrere shows and discusses them well in his work.
The following charts show how the remaining reserves have evolved historically. The start in 1974 is because of the recording and reporting methods, not because that’s is when production suddenly started. Similarly the jump down at 2008 /2009 is, I think, because of changes in SPE reporting methods for reserves introduced in 2007.
Produced and remaining reserves and production history are given in the following two figures: as of December 2015 oil reserves remaining are 3.5 Gb and gas 1.3 Gb. Note the reserves are reported according to SEC rules so they can be developed or undeveloped provided there is a development plan that will allow production within five years. There is a relatively large volume of resources that will be developed but are not included yet (see further below).
Most of the remaining oil is in fairly recent discoveries. Annual production and active drilling rigs are shown for recent years where break out GoM data has been available. I fitted three cycle Verhulst equations to both the oil and gas for discoveries and production. The fits are pretty good except for oil discoveries, which I think are heavily skewed by a couple of really high years. Hubbert logistics curves are symmetric versions of Verhulst equation. To some extent they are just convenient curves that start low have a hump in the middle and then tail off – like production and discovery profiles, But there is some mathematical basis in that they are derived from population growth, so that the production (or discovery) irate is a function of the oil remaining to be produced (or discovered) times that so far produced, which is a kind of proxy for the number of existing facilities or rigs, and that kind of makes sense. The three-cycle agreement matches the shallow, deep and deeper discovery cycles (not all that close though). I don’t know how the skewedness from the non-symmetric curves might be explained in actual physical things, but am looking into it. Hubbert curves from production are used to predict ultimate recovery and peaks, but here the peaks are obvious, and the discovery known.
The fits for production could be used to project the future, but I prefer a bottom up method as below, if data is available, and certainly for oil the Verhulst fit does not do a very good predictive job. Obviously the fit does not predict the sharp increase that occurred through 2016 and is just about peaking now (it is declining while actual production is sharply rising). It would suggest there would have been a peak around 2010 or before if there hadn’t been the drop from the 2008 recession and then drilling hiatus caused by the Deep Water Horizon explosion, which might be true, but isn’t much use for prediction. Short term boom-and-bust upsets like that make curve fitting quite difficult, it would probable need another two or three cycles to get a good match, at which point any pretense that there is some physical basis to the fitting has gone.
The remaining gas is looking pretty sparse, mostly associated gas, which is therefore controlled by oil production and development. Only Hadrian South and Otis are significant recent gas fields. I haven’t seen news of offshore pipelines and onshore gas treatment plants being decommissioned but presumably something like that must be happening or on the cards.
As well as the number dropping the average discovery size has been falling slightly after jumps for oil when deep and ultra-deep fields were first explored. The distribution of discoveries recently has been for few larger discoveries and a number of smaller ones, so a rising proportion of recent developments have been tie-backs. Sometimes these are to new “hub-and-spoke” facilities that just collect several small fields (e.g. Delta House, Na Kika, maybe Chevron’s Tigris in the future), but often the tie-backs are to mature facilities and don’t add to overall production, rather they just extend an existing facility’s plateau production period or reduce the decline rate. However even the smaller discoveries have been drying up over the last couple of years.
The next two figures show the remaining reserves and depletion rates (i.e. percent remaining removed per year at the expected production rate) for the larger or more recently developed fields and for which production data is given later. The rate used in the calculation is taken as the oil production average for the last six months for which there is data. This is therefore a bit out of step with the reserve data – i.e. the production is about 6 to 9 months later than the reserve data, but it’s OK for a rough indication. Note the open bars are used for Stampede and Big Foot as they are not in production but due next year.
Some fields are predominantly gas and the depletion rate counted against oil is not meaningful and probably changes significantly: i.e. Otis, Longhorn, Hadrian South, Mad Dog, Baldpate and Salsa, and partly Na Kika. Some other high depletion rates may be because not all the reserves have been included. For instance Baldpate includes Salsa, Cougar, Deep Penn State and Enchilada, but the last three of those fields do not have reserves listed. Deep Penn State came on line this year, but doesn’t appear anywhere in BOEM numbers (the production is against a combined Baldpate and Salsa lease number). Devil’s Tower numbers may be similar in that Kodiak was bought on line as a new lease but may not show in earlier reserve numbers.
Coelacanth and Son of Bluto 2 have ridiculously low reserves for the size of development, I don’t know if they have been written down from early production data, or maybe the operator was initially very conservative (however early production for Son of Bluto 2 wasn’t very good). Marmalard may reflect missing reserve data. Cardamom doesn’t look to have performed well with high water cut, which might have knocked down the recoverable estimate. Blind Faith has very low production for the available reserves: one lease has been offline for several years and the other is declining fast; the reserves may be overestimated. Overall the depletion numbers should be taken with large error bars; hopefully the next release of 2016 data may sort out a few of these issues.
For all the effort needed to get these fields on line some of the reserves are pretty low – e.g. for Julia and Stones, but it might be there are phase 2 projects planned which are more than five years out and therefore wouldn’t be included – Julia certainly has one such.
Now we should consider EIA reserve figures. They have “proved” reserves from the end of 2015 (C&C) of 4.27 Gb, with change over the year of 10 mmbbls for revisions and adjustments, 108 acquisitions and discoveries, and a drop of 557 for production. They define proved as “… reserves are estimated volumes of hydrocarbon resources that analysis of geologic and engineering data demonstrates with reasonable certainty [defined as 90%] are recoverable under existing economic and operating conditions.” So they would likely include discoveries that might be developed more than five years out. They don’t give field-by-field data so it’s difficult to go much further than that, but see below for further discussion. Over the last two years have been some big potential fields written off (Logan, Kaskida, Hadrian North, Moccasin) and others downgraded (Hopkins, Shenandoah), it will be interesting to see if these have been included in the EIA numbers as their annual update is released; as well as what impact oil price might have had in reserve revisions.
Part II Current Production
The following figures present the BOEM production data for the fields that are shown in depletion estimates above. Some fields have been quite unsteady, which may contribute to some of the anomalies above (though it might be expected that the depletion rates would be shown as quite low if the wells are offline). The results are through April though a few numbers are missing and have been estimated (just by assuming constant daily flow from the previously known month).
The first chart shows oil from fields with leases that have started up since late 2014, followed by their water production. Strictly speaking Mars-Ursa started in the 1990’s but with a second development to the Olympus spar starting early 2014. However it took a bit of time to ramp up and has had a couple of new tie-backs added since, so I’ve included it here.
The average water cut has held steady as more fields have been added, but it would be reasonable to expect it now to increase, as there are few new fields now due. This would usually be accompanied by declining oil production. Some of the fields look to be declining without water support (either from injection or an aquifer), presumably these have solution gas and/or compaction drive, and just go into continuous decline. I do not have the data to be able to identify the drive strategy for each field, though it probably can be found with a bit of searching. The recently announced Kaikias tie-back will go to the Olympus Spar with Mars-Ursa, so some decline must be expected there to make room. Heidelberg, Julia and Stones are still not at their design capacity. Stones has been slow to ramp up and has about 10,000 bpd still to go to meet design. Julia and Heidelberg may have another 40,000 combined but this may wait for later, second phase developments, possibly dependent on initial production results. There has been a phase II development for Jack that will show up in the next few reports through June. Coelacanth is about 25,000 bpd below capacity, but was designed to allow for future tie-backs, so it’s not clear how much of the spare can be filled from current on-line fields. Dalmatian South looks like a waste of money, but there are plans to add multiphase pumping on the Dalmatian fields next year that may have an impact.
Note that a numbers after the name shows how many drilling rigs are currently active on the field.
The next two charts show oil and water for BP operated fields. Note Tubular Bells is a new field included in the chart above. Na Kika is a collection of production of many small and dispersed oil and gas fields (hence the name which refers to an octopus god who created the land out of the sea). Although I’ve collected fields together elsewhere this is the only place I’ve used the facility name for the aggregate as the list of all the fields, e.g. with separators, is too long.
Thunder Horse South came on line early in 2017 and should show in the results for February and March, which have not yet been reported (BP is generally slower than other’s, except sometimes Chevron, with their numbers). I don’t know in which lease it would appear.
The next two figures show oil and water production for Caesar/Tonga/Tahiti, which are blocks in Green Canyon. Three leases are operated by Chevron and go to Tahiti, which was Chevron’s flagship before Jack. Anadarko operates the other four, and its production goes through the Constitution spar, which also processes the Ticonderoga and Constitution fields. It has nameplate capacity of 40,000 bpd and currently looks to be at or slightly over this. There have been two recent fires with shut down and muster on the spar in late March and late April, which might impact production figures. Next year the Constellation field (which was Hopkins when BP operated it) is due to be tied in, so either they expect some decline on existing fields or have done some significant brownfield debottlenecking. For both operators production has been fairly flat and even slightly increasing overall, it looks like gas handling might be the main concern there rather than water – maybe something to look into after the next issue of production numbers.
The final production numbers are for the remaining fields with stated oil reserves above 10 mmbbls. There was a general, and fairly high, decline through 2015 and early 2016, which was arrested after the 2016 hurricane dip. New production came into Shenzi, K2, Baldpate and Devil’s Tower. The Devil’s Tower increase came from a new lease, Kodiak, also shown in the first production chart above. Baldpate tied in Deep Penn South – there is no individual data for this that I can find – and added one other production well. For Shenzi it looks like development wells were added in May 2016 and January 2017, the latter added about 8,000bpd. I think I’m missing something for Shenzi as it has much higher flow than warranted by its stated reserves. BHP, which operates Shenzi, announced neighbouring discoveries at Shenzi North in 2015 and Caicos in 2016, so maybe they just tied them in immediately. K2 added about 12,000 bpd in June 2016, as two leases came back on line and the third increase flow significantly due to the addition of gas lift at the subsea templates.
Tobago includes the Silvertip field. I think there was a plan for a subsea tieback from there in 2016, but the general decline in production suggests maybe not. They are both processed with the Great White field on the Perdido Spar (Shell operator, start-up in 2010). Great White (shown in the chart for new leases) has been increasing, but development drilling was planned to be complete in 2016, so it may now join Tobago in decline.
Of particular note in April: overall there was a big production drop, mainly from Cardamom, Great White and Kodiak going offline; there was a large (five fold) increase in reported water on one Na Kika lease. This may be a reporting revision rather than reality; most lease reports are up to date except Coelacanth, Tubular Bells, Atlantis, the Thunder Horse complex seems to be missing February and March data but have April numbers, it’s not evident that Thunder Horse South start-up has yet added much production yet; Stones ramp up isn’t going well; some of the other small recent new fields aren’t looking great either (Dalmatian, Otis, and Odd Job, which has just started cutting water); Lucius reversed recent declines but the water cut is still increasing rapidly; Julia and Heidelberg are steady but well off their nameplate capacity. Julia and Stones have active drilling rigs, Heidelberg doesn’t. Cardamom goes to the Augur platform, which is well over twenty years old. I don’t know if that might reduce their availability through equipment failures or the need for more planned maintenance – usually older platforms are at low flow and have plenty of on-line redundancy, but Cardamom is a significant proportion of the facility’s nameplate value.
Of the 40 rigs listed by BSEE in the GoM for the past week, thirteen are running tools (some may be exploration wells), four are drilling exploration wells, two are appraisal wells (Phobos and North Platte), two are pre-drilling on Stampede, one each are pre-drilling on Appomattox and Mad Dog (or might be further appraisal) and the others are working on development wells. Eleven rigs are for Shell, five Chevron, four Anadarko, two BP, three Hess and three ExxonMobil.
Altogether the four oil production charts here cover about 90% of total deep water production. There are a large number of other smaller producers in deep plus over 200,000 bpd still from shallow fields. Production from these other fields has actually been holding up pretty well over the last couple of years.
Overall there has been a recent plateau in production from the large mature fields: without the added production to arrest the decline in these fields the GoM would not now be exceeding previous production rates, even with all the new fields that came on line in 2015 and 2016. Some of this will be due to continuous development drilling within a field and other to new tie-backs; facility bottlenecking may also be involved (e.g. K2). Much of this is likely due to investment decisions made in high price era of 2013 and 2014.
With high depletion and low discoveries, and the current investment hiatus, then at some time decline has to start again, and will likely look like 2015 in the chart above. If planned maintenance has been cut back then future decline may actually be worse. The oldest fields are obviously limited by well and reservoir performance, but for some of the middle aged ones, especially those with throughput close to nameplate capacity, the surface facilities may be or are becoming limits. Typically facilities are designed for maximum total liquids at around 50% water cut after that either the oil production falls naturally as the wells water out, or they have to be cut back to allow the production facilities to function correctly (i.e. to stay on line and produce on-spec oil and water), but that’s not hard and fast. Sometimes gas compression will present a lower limit.
The biggest operators in the GoM, and therefore those most likely to influence production in near and medium future, with their average production from January to May in kbpd, are as follows:
|Shell Offshore Inc.||232|
|BP Exploration & Production Inc.||189|
|Anadarko Petroleum Corporation||159|
|LLOG Exploration Offshore, L.L.C.||77|
|Union Oil Company of California||57|
|Chevron U.S.A. Inc||50|
|BHP Billiton Petroleum (GOM) Inc.||48|
(Union Oil is a subsidiary of Chevron.)
Part III Future Scenarios
The following chart presents an educated guess at a scenario for future production, comprising all fields that are identified as probable developments plus estimated decline rates for existing fields based on their expected depletion rates. The total recovery from developed fields, excluding Stampede and Big Foot, is 4.4 Gb, with another 3.5 Gb from new fields. So, given there has been about 0.6 produced in 2016, it may be an overestimate, but equally there may be some additional near field developments that are not yet listed as reserve numbers, or straight reserve growth (e.g. BP recently added over 200 mmboe through improved seismic methods). Note that any of the putative 3.5 Gb additions is currently “undiscovered” it would get added to the cumulative discovery curve against the year of discovery (i.e. the first successful exploration well) when it is approved for development, there isn’t a sudden blip against, say, 2019, should oil prices increase and a number of projects suddenly get approved. However a lot of the recent discoveries are in deep water explored since 2010 so the third cycle in the three cycles of curves mentioned above would become more obvious. These projected recovery numbers are fairly well in line with the EIA reserve numbers, i.e. 7.3 2P compared with 4.3 Gb 1P from EIA, so maybe another 3 Gb of probable, which would reasonable. There are other, I think generally pretty small, ‘discoveries’ that could be added (Sicily, Winter, Samurai, Tortuga, Magellan), but equally some of the fields shown may prove non-commercial.
There may be additional discoveries but with recent discovery size around 10 to 20 mmbbls there’d need to be a lot to make much difference, and therefore a lot of looking, and that isn’t happening at the moment, or opening up a new frontier somewhere. The resources required to meet the start-up profile shown would be extensive, and might not be available in the period shown, especially as the profile implies a sudden price increase and hence completion from other areas to develop delayed projects.
The bracketed number against the field name is the nominal nameplate oil capacity for any development. I’ve also included the curve fit for data through 2015 (black line) – obviously it’s a poor prediction from the start as it misses the 2017 peak.
EIA also provide future projection. Recently these have pretty much consisted of extending the current month’s production for about six months, putting in a dip down for hurricane season if appropriate, and then adding a growth curve of about 300,000 bpd over eighteen months. They have never correctly explained where this extra production comes from – and change their theory with each release, and seem never to allow for any mature field declines. Any largish fields due to produce extra production in 2018 should show in current BOEM reserve data – and only Stamped and Big Foot (which is due for installation late in 2018 and unlikely to be on line) do. I think reality will be closer to my scenario: allowing for new discoveries and increased brownfield and in-fill activity it might be nearer a plateau than shown in the short term, but there will not be growth before, at minimum, 2021 and then only if there are major discoveries.
The GoM is a mature basin, the deep-water fields less so than shallow, but it looks currently that oil is in slow terminal decline and gas mostly exhausted. Many other offshore basins (and probably onshore) are in a similar state or soon will be: UK, Norway, Mexico, Azerbaijan, most Asian countries and Australia. Less so for Angola and Nigeria but the current hiatus in investment will hit their production soon – likely their final production profile will look like a two or three cycle Verhulst fit, a bit like the GoM oil, with a coming dip. It has good data availability, even if the basis using lease numbers can be confusing, therefore it is quite interesting as an indicator of what happening in other, less well documented, basins. I also find it amusing to follow the knots that the EIA, the industry media and the E&Ps tie themselves into in order not to openly admit there might ever be a peak.