Cassandra has moved. Ugo Bardi publishes now on a new site called "The Seneca Effect."

Sunday, December 16, 2018

Peak Diesel or no Peak Diesel? The Debate is Ongoing

In a recent post, Antonio Turiel proposed that the global peak of diesel fuel production was reached three years ago, in 2018. Turiel's idea is especially interesting since it takes into account the fact that what we call "oil" is actually a wide variety of liquids of different characteristics. The current boom of the extraction of tight oil (known also as "shale oil") in the United States has avoided, so far, the decline of the total volume of oil produced worldwide ("peak oil").

Shale oil has changed a lot of things in the oil industry, but it couldn't avoid the decline of conventional oil. That, in turn, had consequences: shale oil is light oil, not easily converted to the kind of fuel (diesel) which is the most important transportation fuel, nowadays. That seems to have forced the oil industry into converting more and more "heavy" oil into diesel fuel but, even so, diesel fuel is becoming gradually more scarce and more expensive, to the point that its production may have peaked in 2015. In addition, it has created a dearth of heavy oil, the fuel of choice for marine transportation. In short, the famed "peak oil" is arriving not all together, but piecemeal -- affecting some kinds of fuels faster than others.

Turiel's proposal has raised a considerable debate among the experts, with several of them challenging Turiel's interpretation. Turiel himself and Gail Tverberg (of the "our finite world" blog) discussed the validity of the data and their meaning. Below, I reproduce the exchange with their kind permission. As you will see, the matter is complex and at the present stage it is not possible to arrive at a definitive conclusion. In my personal opinion, I would say that it is understandable that many of us are afraid of being criticized for having called wolf too early, but that it is nevertheless worth reporting one's data and discuss them on the basis of what we know. Then, as attributed to John Maynard Keynes, "When I have new data, I modify my conclusions. What do you do, sir?"

Gail Tverberg

Dear Ugo,

I don’t know if you have noticed, but data by type of refined fuel is available from various standard sources of energy data. EIA data has a lot of detail data for the US; BP has regional data for a number of breakdowns. There are no doubt other sources for oil consumption by country. I think of JODI as voluntary data; it is not really clear (to me) which countries are in or out, for which periods.

The information you are showing in your recent post seems to show a fairly different pattern from what BP shows (Dist. means Distillates).

According to BP, Middle distillates consist of jet, heating kerosenes, gas and diesel oils (including marine bunkers).

Within Medium Distillates, there is a further breakdown for recent years, showing a category called diesel/gasoil separately from jet/kerosene. It shows a fairly similar pattern.

It is the “fuel oil” category, which seems to be the heavy distillates, that shows the big downturn in consumption. This is consistent with what we see in the US. Refineries can make a lot more money if they crack heavy oil and refine it into lighter products than if they sell it in close to the unrefined state. In the US, much road construction has changed from asphalt to concrete. Concrete is a coal product in some parts of the world.

In the US, petroleum coke has also shown a big downturn.

With respect to what EIA calls distillate fuel oil (which I think of as diesel), in the US, there indeed were two big steps down.

The first downturn in consumption, in 1981 (when interest rates were raised), was when a lot of home heating and also electricity generation was switched from diesel to other energy products. The second downturn occurred in 2008, when even more homeowners switched away from using diesel for home heating. Also, on the industrial side, some new techniques were developed for drilling oil wells, using natural gas instead of diesel. Natural gas is usually produced in the same field, and is much cheaper for oil producers to use, rather than purchasing diesel. Note that the percentage downturn is far smaller in the "distillate fuel oil” chart than for the other two EIA charts I showed.

To me, it is very difficult to figure out exactly what is happening, with such similar names for different products. Also, there seems to be a lot of shifting of use around the globe. All of this makes the situation confusing.

You might want to backtrack a bit on what you said about diesel. The evidence doesn’t seem as strong, looking at other sources. Perhaps a different post, looking at some new data as well, would be in order. BP data can be downloaded from this link: The tab you are interested in is Oil - Regional Consumption.

Best wishes,


Antonio Turiel

Dear Gail,

As for any other peak, some years must be spent to be completely sure that we have passed them. So any evidence so far must be taken, always, with a bit of caution, and in that sense I agree with your warning.

Regarding data, I prefer to use JODI because JODI data is better grounded than EIA data - EIA data contains a lot of "inference", typically spawning from six months in the US up to a couple of years in other countries. JODI data, on the contrary, tend to be more timely (and when there are significant time lags these are reported). Notice also that EIA, IEA and BP use JODI data as one of their sources.

Another significant difference is that in your first graph you represent consumption, while I always represent production. The difference is significant, as I am mainly interested in the refinery throughput because this is the problem I want to characterize (the difficulties to increase the output). The use of stored stuff explains the difference between both.

Some readers have pointed out that the slowdown and even decrease in diesel production, if real, could be a consequence of a lowering demand. This is the same situation as for peak oil: you can always argue that there is not enough demand for that oil, and it is true in any instance: the problem is one of affordability, as you have explained many times.



Gail Tverberg
Hello Antonio,

I think that there is a real difference between the kind of data a person wants to look at when that person is examining the indications for an individual country or a subdivision within a country and the information a person wants to look at for world level indications.

When a person is looking at detail level data, then I agree that there is very often a big difference between production and consumption. Looking at data such as JODI data, along with other indications, can be helpful for putting together the true indications for that small grouping. A person has to be pretty aware of particular patterns for individual countries or other smaller groupings. I know, for example, that Texas shale oil data seems to be reported much more slowly than North Dakota shale oil data. Some countries are notorious for trying to exaggerate their production. This is why OPEC shows two sets of numbers, in its monthly reports: “from secondary sources” and “as reported by the producer.” The “from secondary sources” numbers are generally viewed as the more accurate ones.

When a person is looking at small segments of data, corresponding more or less to how the data is reported, then it is fairly easy to see major mistakes. For example, does it look like the "diesel” (or some other grouping) accidentally got reported as “fuel oil,” for some period of time? Does it look like some categories are simply missing, or the amounts have been misinterpreted? If I am looking at detail data, then I can look for mistakes. By the time aggregations occur, the big problems, like missing whole sets of data from some small countries, will be difficult to see. If I am looking at aggregate data, especially on a world basis, I really want someone to have looked at the data in detail, and to have figured out what pieces were missing. They have no doubt made some estimates of the missing pieces, but if I am making estimates of trends, making estimates of the missing pieces is absolutely essential.

I personally have no experience working with JODI, but I have worked with an awfully lot of other data sets (in the insurance world previously, and now in the energy world). I am very much aware of the fact that the initial coding is likely to have a lot of flaws, especially if it is voluntary, and doesn’t have to balance to published financial data.

There is indeed some difference between production and consumption, but when we get to a world basis, they mostly offset. For the purpose of determining trends, what we want is well-massaged data–data that is as free from errors and omissions as possible. I would be willing to believe EIA, IEA, or BP data for this purpose. I would much prefer using well-massaged consumption data to look at trends, rather than a summation of individually reported data of questionable validity.

I have at least a little background on what is happening. I know that there is a fair amount of flexibility in the distribution of finished oil products that can be obtained from a barrel of oil. In general, it is possible to “crack” long hydrocarbons to make shorter (and thus lighter) hydrocarbons; it is close to impossible to go from short chains to longer chains. I was involved in discussions in 2008, when oil companies wanted to increase the refining of what had been products such as asphalt and petroleum coke, because, with high oil prices, oil companies could make a much larger profit from refining heavy oil into higher-price products such as diesel and gasoline. Concrete could be substituted for the asphalt. The US has a natural advantage in cracking long molecules because it has an abundant supply of low-cost natural gas. That keeps the cost below what a similar process would cost in Europe. Heavy oil, such as that from the oil sands, also tends to sell at a substantially lower price than light sweet oil, making the process profitable in the US under a range of price scenarios.

When I see two different trends, one in the JODI data and a different one in the BP data, I am inclined to believe the BP indications.

A Different Diesel Problem

I think that Europe may have a different diesel problem than the one you are thinking about. Europe has tended to use diesel to power its private passenger automobiles as well as its trucks. This is an awfully lot of “demand” to put on one segment of refined products from a barrel of oil. The US and many other countries have spread out demand, with private passenger automobiles using gasoline, instead of diesel. This allows for demand to match up better with what comes out of a barrel of crude oil. According to BP data, in 2017, Europe consumed 7.7% of the world’s gasoline supply and 24.4% of the world’s supply of a subcategory it calls diesel/gasoil. (These are subcategories for recent years that I don’t show on the chart above.) I suspect that there is no oil, anywhere, that could be refined to provide the overly heavy diesel mix that Europe requires. No one in Europe stopped to think, “If cars and trucks both run on diesel, we will need to import an awfully lot of diesel from the world market. We are asking for problems. If the world has barely enough to go around, our demand will raise world diesel prices.”

At this point, there is no sense in adding a whole lot of refining capability for heavy oil in Europe; Europe lacks the cheap natural gas to process it. The same BP report mentioned previously also shows data on Europe’s refining capacity and its refinery throughput. Refining capacity and throughput both seem to be falling, as available North Sea oil falls.

Best wishes,

Gail Tverberg

Antonio Turiel

Dear Gail,

Sorry for my late response - I'm presently attending an important conference in Rome, and the previous days I was very busy preparing my presentation.

Regarding your comments, if I understand correctly your point, you prefer EIA, AIE and BP data as they have better quality, apart from the fact that they integrate diverse sources of data. The key point is that they apply a better quality control and the result is, let's say, better.

This is a reasonable point, but something that I anyway call into question: are those data really better? As a matter of fact, both EIA and IEA suffer political pressures to make up their data, and this kind of thing is much worse than having an error: it is a bias. Random errors (unexpected data failures, data flow interruptions, occasional double accounting, etc) do not really change the trend, just increase data volatility, something that can be compensated by for instance averaging (e.g., the sliding window of 12 months we apply). But biases can change trends, and that's is quite crucial.

Your point is that maybe what is called diesel has changed along JODI series, something that I am not absolutely aware of, and in fact such a "sudden removal" of diesel from that category should result in an increase the other middle distillates, the "Other fuel oil" category, which is not the case. Besides, removal as such typically shows up as steps in the graph, something that is not observed either. So such hypothesis seems to me very unlikely.

Coping with noisy data with unknown uncertainties is something physicists are used to do, because this is our bread and butter (data from the real world are always noisy and uncertain).

So let me tell you what I propose to solve this issue:

I'm a specialist in a technique called "Triple Collocation" that allows an intrinsic characterization of errors and biases of three sets of different measurements of the same physical quantity. Therefore, abusing of your kindness, if you could provide me different data series of data of what you could name "diesel" or "medium distillates" or whatever you feel more confident of (or even better, all of them!), from different data providers you trust the most (EIA, IEA, BP, whatever) and I will include the data from JODI and make all possible triples (if we have EIA, IEA, BP and JODI we have 4 possible triplets), I can estimate the calibration factors, biases and standard deviations of the random errors for each triplet, then compare the 4 possibilities to see if the results are consistent.

This exercise could be very informative for all us and provide a better insight about where we actually are right now.




  1. Dear Doctor Bardy.

    An authentic duel of titans, one of the best post of the year.

    Please, I want to see the end.

  2. I'm curious as to when everyone here thinks Peak Shale will kick in. I read somewhere that the production of new wells drilled in the third quarter of 1918 was less than for the same period in 2017. That sounds like most of the "sweet spots" have already been drilled and they will have to be drilling even more desperately in less productive areas just to maintain production.

    It seems that when we hit Peak Shale, we will also have hit Peak All Liquids. Whatever our short term concerns with diesel may be, I think the real blow to the economy will come within a year or two of Peak All Liquids. If I recall aright, it was Art Berman, some months ago that gave two to three years before Peak Shale. Does this sound about right to everyone here, and also does it seem like my idea of Peak Shale and Peak All Liquids happening about the same time seem likely.

    Antoinetta III

    1. The zillion dollar question, Antonietta. And, as a descendant (I imagine) of Queen Marie Antoinette, once the peak arrives, you could say "let them eat palm oil"

      (BTW, the queen never said let them eat cake, of course)

  3. Thanks for that Ugo, looking for the sequel !
    One thing for sure EU27 has passed its "peak oi consumption"(demand or whatever one wants to call it), in 2005 or something like that, for instance :

  4. In the case of petroleum coke, or petcoke, as it is commonly called, this is a byproduct of processing asphalt or other very heavy residual oils into as much middle and/or light distillate as possible. The asphalt is heated to a very high temperature in a coking unit to break down heavier hydrocarbons (pyrolysis), and vapors driven off by this process can be cracked, hydrotreated, etc. into liquid fuels. When this first became common practice in the US several years ago, the price of asphalt increased quite a bit as so much of it was diverted into making fuels. I am not a petroleum engineer, but my guess is that if less petcoke is being produced, the process has improved over the years to leave less coke when complete. As I understand it, refining extra-heavy oils, such as Canadian bitumen and Orinico, is not practical without coking, unless the objective is to produce a large proportion of asphalt. (Note that some buyers in China are taking shipments of Canadian bitumen for this purpose.) The US has been buying large amounts of Canadian crude for the past several years, and I actually found the reduction in produced petcoke surprising.

  5. The general trend in the US is toward producing more petcoke.
    But the amount of oil imported from Canada, especially the increase compared to several years ago, is primarily extra-heavy.
    And the drop in imports from Venezuela is less than the increase from Canada.

  6. I have a bit of an issue with Gails statement that 2008 drop was due to less heating oil demand.

    2005 to 2008 I lived in the middle of the Willamette valley OR. I had a restaurant and was working closely with area farming. During that period diesel fuel was on a steady climb in price. One of my suppliers, Sysco Foods first sent out a notice that they were going to absorb cost as it was just a blip. Six months later they said they were going to increase price on certain items that were highly effected by the fuel increase. Later they notified everyone that there would be a fuel charge on all orders, then they sent out a notice warning restaurant owners to increase their prices as these increases were not going away and some restaurants were going broke trying to absorb the cost and not pass it on to customers.

    During all this what they call offroad diesel or farm diesel, the red stuff higher in sulphur, was going up and up and becoming unavailable in some areas. BIG problem. Also nearly all farm inputs were increasing dramatically, Ammonia went from $100 per ton to almost $1000. Around here there were as many or more diesel pickups as gas, after a couple years you could not give a diesel truck away and compact cars lined all the parking areas on the farms and job sites.

    I don't doubt that a lot of households moved away from heating oil during this period but that is only Incidental.

    I do remember reading some analysis pointing to a booming EU economy and rising diesel demand there and around the world as at least part of the puzzle. With that in mind it would seem that due to the "Global economy contraction" and EU down turn, it would seem that peak diesel has been given a bit of a temporary reprieve, which is how peak oil seems to be playing out in general.

  7. Aviation fuel is pretty close to diesel, isn't it? It will become available for vehicles soon, when no-one wants to fly anymore due to peer pressure. Flyers are terrorist massmurderers, you know...

  8. There appears to be an error with Turiel's data. I compute 2017 [2015] average (not 12 month trailing average) gas/diesel oil refinery output from the JODI - Secondary (all data) set as 26.275 [26.308] Mb/d. Just rounding errors.

  9. Diesel is needed to bring diesel to us, diesel fuel is needed by the Oil Industry. Oil industry gets first pick , and the global economy gets what is left over. Any further fall in EROI for diesel compounds on the global economy, on mining, and oil EROI to the end point of exergy release.
    Diesel flat-lining is a sign of less net available diesel, and real global economic decline. Demand fall and changes in production allocation, and real financial and economic stress are all happening at the same time, and the workers real wage declines and falls in consumption. Its all linked and now maybe too hard to disentangle cause and effect as everything starts to change and adapt at the same time. Maybe flat-lining diesel is a sign of this systems real Seneca Collapse stage in progress.

  10. Others have looked at Turiel's JODI data. It's a nothingburger and the last month of data will always be a downleg.

    How anyone can take the JODI data set seriously when Iran is missing 3 months and there is no output from Venezuela in 2017 is beyond me.

  11. Contradictory data sets is a fact of life in reporting or analyzing trends in the energy realm. The reasons for these discrepancies is not always clear. It can be as simple as using different units or in the case of renewable sources using nameplate capacity vs real world delivered capacity. There is also the Mark Twain issue of "Figures can lie and liars can figure." Another issue not noted(?) is whether the anticipated switch of ocean carriers using bunker fuel to using #2 diesel will draw from the #2 diesel pool used by trucking , Ag and heavy industry thereby inducing scarcity and possible price jumps. It would seem to me that increasing middle distillate prices say for heating oil will cause those customers to switch if they can to a cheaper heating source and thereby free up that fuel which could flow to #2 diesel users. For example if diesel became twice as expensive as gasoline some sectors might be able to use gasoline to run their trucks or tractors but that would only happen if the gas/diesel differential stayed wide and prolonged. Big Trucks or tractors in my field are a rarity and expecting engine makers to change back to gasoline power would probably not happen rapidly enough to make a difference. Conventional crude is the ideal feedstock for diesel production and that is declining and if that continues diesel production will follow unavoidably. I eagerly awaid Ugo and Gail and Antonio continuing this thread. Is diesel production really declining or is it a statistical aberration? Terrific post UGO!

    1. Keep in touch, sv. I have some data that indirectly support Antonio's idea. Hugely interesting data, I'll publish the story after Christmas

  12. Ugo, Gail,
    Another caution for following energy supply trends is failing to allow for the tremendous creativity of the global economy. I see that displayed in the remarkably regular exponential shapes of the global energy use totals and related factors. JPEG -,M,F,E,CO2,Eff-1971-16Large.jpg

    As a set of closely related curves with constant growth rates they clearly define a family of current natural growth constants for the world economy. Why that is, and why it has not been discussed much, are a considerable mystery, of course. It's strange that the global economy appears to work so very smoothly, even while everyone's perception is of things fluctuating quite unpredictably all the time.

    I think that contradiction is itself the clue to the real answer. What the smooth global data would seem to display is the economy working as it is supposed to, actively and creatively allocating needed resources to where they will be most productive. For energy, it means the economic markets are finding and developing alternate energy resources, as needed, to satisfy the continually growing demand for global energy supplies.

    So I think as long as maximizing the global rate of economic growth is the institutional design of the economy, and common driving purpose of both the financial and political worlds, growing competitive energy resources and will continue to be supplied. At present, we seem to have the technology and creativity to discover new sources as needed, so I expect it will continue for now at least, to fill in whatever gaps develop.

    Of course, that's terrible for climate mitigation efforts; having a purpose in direct conflict with the current economic organization and motivation for globally maximizing the rate of growth. So perhaps our interests in recognizing approaching peaks of overdevelopment and the hazards that come with them may only need to creatively follow the threads to focus on the right subjects and scale for exposing the questions people need to face.

    All the best,



Ugo Bardi is a member of the Club of Rome, faculty member of the University of Florence, and the author of "Extracted" (Chelsea Green 2014), "The Seneca Effect" (Springer 2017), and Before the Collapse (Springer 2019)