The growth of renewables has been revolutionary, with wind and solar emerging towards the bottom of the global cost curve, scaling up at a pace of 270TWH pa. However, we find unsettling evidence that the market could slow by c15% from 2020, plateauing in heartland geographies such as California, Germany and the UK. The rationale, and all the underlying data, are included in this 6-page PDF research report and associated Excel file.
This short note outlines our top conclusions about the energy consumption of the internet, which now comprises c2% of global electricity and 0.7% of global CO2. In the next decade, remarkably, the CO2 footprint of powering the internet could surpass that of producing oil or gas.
We argue CO2-labelling is the most important policy-measure that can be taken to accelerate the energy transition: making products’ CO2-intensities visible, so they can sway purchasing decisions. There is precedent to expect 4-8% savings across global energy use, which will lower the net global costs of decarbonisation by $200-400bn pa. Digital technologies also support wider eco-labelling compared with the past. Leading companies are preparing their businesses.
The key challenge for the US shale industry is to continue improving productivity per well, as illustrated repeatedly in our research. Hence, this short note reviews an advance in fracturing fluids, which has been patented by BP. Diverter compositions are optimised across successive pressurization cycles, to create dendritic fracture geometries, which will enhance stimulated rock volumes.
Precision-engineered proteins are on the cusp of disrupting the meat industry, according to an exceptional, 75-page report, published recently by RethinkX. The science is rapidly improving, to create foods with vastly superior nutrition, superior taste and superior costs, by the early-2020s.
The energy opportunities are most exciting to us, after reading the report. If RethinkX’s scenarios play out, we estimate: direct CO2 savings of 400MTpa, enough to offset 10% of US oil demand; 2bcfd of upside to US gas demand; and enough land would be freed up to decarbonise all of US oil demand, or increase US biofuels production by 6x to c6Mbpd.
We would be delighted to introduce clients of Thunder Said Energy to the reports’ authors, Catherine Tubb and Tony Seba. Please contact us if this is useful.
Over 100 attacks on global energy assets made major news headlines in the past decade. The majority were small-scale, targeting pipelines in conflict-regions, because this was the infrastructure most accessible to aggressors. However, a new and devastating wave of drone technologies could place the world’s largest and most vulnerable facilities into the firing line, threatening multiple millions of barrels per day. This short note outlines the latest in drone technologies and why they concern us.
Historical attacks on energy assets
Supply disruptions have been a feature of oil markets over the past ten years. For example, in the chart below, we have counted 100 violent attacks on energy infrastructure from major news stories. However, the majority were small-scale and located in active conflict-zones. Most oil infrastructure has heretofore been safe.
Here are the numbers: 90% of the prior attacks in our sample were low impact, when we assessed their severity. c60% were concentrated on pipeline infrastructure, which is relatively easy to repair. 70% of the upstream attacks were on wells or small processing units. 80% were localised within active war-zones such as Libya, Nigeria, Iraq, Yemen and the Sudans, rather than in stable countries. These attacks were nevertheless numerous. They shuttered 1Mbpd of Nigerian output between 2006 and 2016, 1Mbpd of Libyan output in 2011 and c0.5Mbpd of output in Yemen and Syria.
The more dangerous and worrying attacks have been full-scale assaults on large industrial assets. The worst example, many will remember, was Al Qaeda’s January-2013 attack on Algeria’s 9bcm pa In Amenas gas facility. 39 hostages were killed, as well as 29 terrorists. In addition, it took until June-2016 to bring production back to full capacity. The impacts of such incidents are hard-felt and long-lasting. Another legacy is that security measures have been escalated in high-risk regions.
On 14th September 2019, another industry-changing attack took place, on Saudi Arabia’s Abqaiq and Khurais oilfields. 5.7Mbpd of oil production was curtailed, constituting the largest supply-disruption on record. Repairing the damage will cost hundreds of millions of dollars. The latest suggestion is that the damage was inflicted by 20 drones, plus additional cruise missiles, which may have been guided to their targets by the drones. Unfortunately, this attack raises the spectre of further incidents, owing to the rise of drone swarm technology.
Ten Characteristics of Drone Swarms
Drone swarms could emerge as the most devastating weapon of 21st century warfare, outflanking large, high-speed, high-cost military vehicles of the past (Hambling, 2015; chart below, data here). They pose much greater risk to high-value infrastructure than prior weaponry that was available to aggressors. To understand why, it is necessary to review ten properties of drone swarms.
(1) Easy to access. Most military equipment is not openly available for purchase on the internet or in consumer electronic stores. However, hundreds of models of drones are now available in the consumer sector. They can be modified and retro-fitted to inflict violence or damage. Similarly, in the military sphere, one expects large super-powers such as the US, Russia and China to develop leading military technologies, but advanced drones are also being developed in smaller countries such as Israel, Iran, Turkey, Korea. The technology is not always closely contained. In particular, Iran has been found to donate its Ababil drones and Quds missiles to allies such as the Houthis; and Islamic State was able to use drones to drop grenades in Northern Iraq in 2016-17.
(2) Easy to fund. These drones have price points in the thousands of dollars, rather than the millions, which makes them accessible to small groups of aggressors rather than just to nation-states. Out of 15 high-spec consumer drones that we reviewed recently, the median cost was $10,000 (chart below, data here). Half-a-dozen priced below $2,500. This not only makes them accessible, compared to cruise missiles costing $150k to $1.5M; but also expendable, compared to fighter jets costing $30-150M.
(3) Easy to launch. There is no need for runways, special hangers or refuelling facilities. Drones can launch from any terrain and travel tens or hundreds of miles. The fact that drones can be launched and travel to their targets brings a much wider array of assets into the firing line. This will include facilities deep within protected territory, such as Abqaiq and Khurais; or offshore assets, which have repeatedly been considered as targets by Nigerian militants, but have been protected by their offshore locations.
(4) Increasingly large swarms. In 2015, the largest drone swarms being flown numbered 30-50. However, China’s CETC flew drone swarms numbering 100-200 in 2018 (chart below). Israel is developing technologies where a single operator could fly an entire swarm of drones, in a single, controllable formation. This matters because the larger the swarm, the harder it is to neutralize. Using a swarm of 20 drones may be one reason why the latest attack on Saudi infrastructure succeeded, while dozens of prior attacks from 2017-18 were thwarted.
(5) Increasingly autonomous swarms. The most effective counter-measure against military drones in the past has been to “jam” the controllers used for steering them. This tactic was used, for example, against Islamic State, in Northern Iraq. But now, some of the leading commercial drones use neural network algorithms to auto-navigate. Thus they cannot be “jammed”. For example, the Skydio R1 uses a NVIDIA Jetson processor with 192 processing cores, which is less power hungry than prior chips. Qualcomm is also making ‘simultaneous location and mapping’ hardware the size of a credit card, allowing drones to navigate by sight alone.
(6) Potency. A large drone may carry a warhead or missile; smaller drones can carry grenades, IEDs or firearms and small drones may illuminate targets (e.g., with lasers) in order to direct larger incoming missiles. Any of these could do very significant damage to facilities that contain live hydrocarbons.
(7) Precision. Autonomous drones can attack very specific targets. This level of precision was seen in the recent Saudi attacks, where individual missiles hit each spheroid tank at Abqaiq, in almost the same identical location (US satellite images below). Another example in the civilian sector is being used at beaches in Australia, where ‘SharkSpotter’ deep learning software is used to identify sharks with 90% accuracy, compared with 30% for human operators. Training a drone to identify sharks versus dolphins is computationally similar to identifying vulnerable versus non-vulnerable processing units at energy infrastructure.
(8) Hard to predict. Because swarms of drones are created with standard electronics equipment, much of it available in the civilian sector, “manufacture [of drone swarms] would be relatively hard to spot—compared to the production of traditional military hardware such as manned aircraft, ships or ballistic missiles—as it would resemble any other consumer electronics assembly” (Hambling, 2018).
(9) Hard to stop. The challenge of stopping a large swarm of drones is that there may simply be too many units to neutralize, especially when they are moving quickly. Laser cannons may stop a few units. A battery of missiles may stop many more. However “shooting down a $1,000 drone with a $5,000 missile is not a winning strategy” (Hambling, 2015). Assuming similar budgets, the drone attackers may outnumber the missile defenders. Acknowledging this challenge, the US has budgeted $1.5bn over the next year, to investigate potential solutions. But outside the military, and back in the realm of energy assets, we doubt that any of today’s onshore or offshore processing facilities have the capacity to stop drone attacks.
(10). Hard to retaliate. Drone attacks are very different from prior cases where armed insurgents attacked oil infrastructure, risking their own lives in the melee. Drones are by their nature remotely operated. Furthermore, reading through the history of recent drone attacks (e.g., in Yemen and Syria), it has often been impossible even to identify the culprit. In some cases, their identity still remains disputed. Failure to pinpoint the perpetrator makes it difficult to strike back. In turn, this removes the usual deterrent to attacking an enemy.
Implications for Oil Markets and Companies
Our latest oil market forecasts point to 1-2Mbpd of over-supply each year in the 2020s, assuming steady demand growth of 1.3Mbpd per annum. However, these base case forecasts do not incorporate any impact of supply disruptions from further attacks, which could sway the balance, and cause significant price spikes.
For energy companies, we think it will be crucial to mitigate against the risk of drone strikes, to the best extent possible. This may include diversification, counter-measures, and a growing preference to operate in lower-risk countries. We would be very happy to introduce clients of Thunder Said Energy to our contacts in the military drone space, who may be able to provide further observations.
What will happen to oil refineries during the energy transition? On our numbers, liquid oil products will be needed past 2100, long after demand plateaus in the 2020s. Cleaner, more efficient technologies are therefore required in the downstream sector. This note considers whether refineries could increasingly be converted to bio-refineries.
Our evidence comes from the patent literature, as we have reviewed 3,000 patents from the leading 25 Energy Majors. 8% are focused on new energies (chart below, full details in our deep-dive note). Eni screens as the leader for converting refineries to bio-refineries, hence this note summarises its relevant patents on the topic.
Historical Context. Use of vegetable oils in diesel engines goes back to Rudolf Diesel, who, in 1900, ran an engine on peanut oil. Palm oil and peanut oil were both used as military diesel in Africa in WWII. However, vegetable fuels were abandoned due to high costs and inconsistent quality, compared with petroleum fuels.
Today’s vegetable oil fuel-blending components primarily contain Fatty Acid Methyl Esters (FAME). However, they cannot be blended beyond c7% without causing problems in auto engines. For example, FAME has a low energy content (38kJ/kg vs diesel at 45kJ/kg), a -5 – 15C cloud point, causes pollution in tanks, polymerises to form rubbers, causes fouling, dirties filters and contaminates lubricants.
Regulation is nevertheless stoking demand for more dio-diesel, going beyond the 7% threshold. Europe Directive 2009/28/C mandates 10% renewable material in diesel by 2020, up from 5% in 2014.
Eni is therefore converting refineries to bio-refineries, to upgrade renewable materials into “green diesel”. A 0.36MTpa facility started up at Porto Marghera, Venice in 2014. A larger, 0.7MTpa facility started at Gela in 2019. Both convert vegetable oils into diesel.
Patents indicate how they work. The starting point is a conventional oil refinery, with two sequential hydro-desulfurization units. For the conversion into a bio-refinery. these units are re-vamped into a hydrodeoxygenation reactor (HDO) and a subsequent hydro-isomerization reactor (ISO), shown in the schematic below.
- HDO occurs in the presence of hydrogen, a sulfided hydrogenation catalyst from Group VIII or VIB metals, at 25-70 bar and 240-450C.
- ISO occurs at 250-450C, 25-70bar and a Metal (Pt, Pd, Ni) Acid catalyst on an alumino-silica zeolite framework.
- Upstream modifications. Pre-treatment processes, surge drums and heat-exchangers are installed upstream of each reactor.
- Downstream modifications. The output products from the reactors will contain 1-5% H2S, which is removed in an acid gas treatment unit, and then a Claus unit for sulphur recovery; both reached via new connection lines.
The main advantage of this process is cost, which is said to be 80% lower than constructing a new facility. For example, the Porto Marghera project was budgeted at €200M. In its patents, Eni states: “This method is of particular interest within the current economic context which envisages a reduction in the demand for oil products and refinery margins”.
Further advantages are that the produced diesel has excellent properties, including a high octane index, optimum cold properties, high calorific value and a further by-product stream of commercial LPGs. Moreover, the efficiency of the converted facility is seen to be similar to one constructed anew.
The disadvantage is that blending of free fatty acids is limited to c20%. This is why the bio-refineries so far intake 80% palm oil (which contain <0.1% free fatty acids). Eni states: “The reactor used for effecting the HDO step, deriving, through the method of the present invention, from a pre-existing hydrodesulfurization unit, may not have a metallurgy suitable for guaranteeing its use in the presence of high concentrations of free fatty acids in the feedstock consisting of a mixture of vegetable oils. The reactors of the HDO/ISO units specifically constructed for this purpose, are in fact made of stainless steel (316 SS, 317 SS), to allow them to treat contents of free fatty acids of up to 20% by weight of the feedstock”. Processing a broader range of vegetable oils and other waste oils would require a more costly refinery re-vamp.
Further challenges are that the production of hydrogen and other industrial above will be energy intensive. Moreover, Eni’s 1MTpa of green diesel production capacity is only equivalent to c20kbpd of fuel. It will be challenging to source sufficient feedstocks to scale bio-refineries up to meet larger portions of the world’s overall fuel needs.
Our conclusion is therefore that bio-refineries have potential when re-purposing existing downstream facilities, preserving value in the very long-term future of the industry. However, further technological improvements are required before these facilities can scale up or deliver material, and truly decarbonised hydrocarbons. Out of Eni’s other refining patents, we are most positive on Eni Slurry Technology, which is a leading technology for IMO2020 (chart below). For details of other technology leaders in energy, please see our note, Patent Leaders.
Source: Rispoli, G., F. & Prati, C. (2018). Method for Re-Vamping a Conventional Mineral Oils Refinery to a Bio-Refinery. US Patent US2018079967.
Gas demand could treble by 2050, gaining traction not just as the world’s cleanest fossil fuel, but also the most economical. The ascent would be driven by technology. Hence this note outlines 200MTpa of potential upside to consensus LNG demand, via de-carbonised power and shipping fuels. LNG demand could thus compound at 8% pa to 800MTpa by 2030, justifying greater investment in unsanctioned LNG projects.
What if there were a technology to sequester CO2, double shale productivity, earn 15-30% IRRs and it was on the cusp of commercialization? Promising momentum is building, at the nexus of decarbonised gas-power and Permian CO2-EOR…
First, this week, we finished reviewing 350 technical papers from the shale industry’s 2019 URTEC conference. The biggest YoY delta is that publications into EOR rose 2.3x. CO2-EOR is favored (chart below). Further insights from the technical literature will follow in a detailed publication, but importantly we do not see underlying productivity growth in shale to be slowing.
Second, we re-read Occidental Petroleum’s 2Q19 conference call. More vocally than ever before, Oxy hinted it could take the pure CO2 from decarbonised power plants and use it for Permian-EOR; with its equity interest in NetPower, 1.6M net Permian acres, and leading CO2-EOR technology. Quotes from the call are below:
- On CO2-EOR: “We are investing in technologies that will not only lower our cost of CO2 for enhanced oil recovery in our Permian conventional reservoirs, but will also bring forward the application of CO2 enhanced oil recovery to shales across the Permian, D.J. and Powder River basins”
- On decarbonised gas power: “What it does is, it takes natural gas combines that with oxygen and burns it together, and that’s what creates electricity and it creates that electricity at lower costs… one of our solutions is to put that in the Permian… for use in our enhanced oil recovery… It will utilize our gas that that if we sold it would make nearly as much”.
- On the opportunity: “We are getting calls from all over the world, with people wanting our help to — figure out how to capture CO2 from industrial sources, and then what to do with it and oil reservoirs”.
Our extensive work on these themes includes two deep-dive reports linked above. Our underlying models can connect c10% IRRs on oxy-combustion gas plants (first chart below) with 15-30% IRRs at Permian CO2-EOR (second chart below). On these numbers, the overall NPV10 of an integrated system could surpass $10bn.
EOR remains one of the most exciting avenues to boost Permian production potential. So far, our shale forecasts assume little direct benefit (chart below). But an indirect benefit is implicit, as we assume 10% annualized productivity growth to 2025, which would underpin a very strong ramp-up (chart below). 2023-25 currently look well-supplied in our oil market model, due to falling decline rates, but this could be compounded by CO2-EOR.
We are more positive on the ascent of gas, stoked by increasing usage in decarbonised power. We see potential for gas demand to treble by 2050.
Technology drives 30-60% of energy companies’ return on capital. This is our conclusion after correlating 10 energy companies’ ROACEs against 3,000 patent filings. Above average technologies are necessary to generate above-average returns.
For the first time, we have been able to test the relationship between oil companies’ technical abilities and their Returns on Average Capital Employed (ROACE).
In the past, technical capabilities have been difficult to quantify, hence this crucial dimension has been overlooked by economic analysis in the energy sector.
Our new methodology stems from our database of 3,043 patents, filed by the Top 25 leading energy companies in 2018. The data cover upstream, downstream, chemicals and new energy technologies (chart below) . All the patents are further summarised, “scored” and classed across 40 sub-categories.
The methodology is to correlate our patent-scores for each company with the ROACE generated by the company in 2018. We ran these correlations at both the corporate level and the segment level…
Results: patent filings predict returns
Patent filings predict corporate returns. In 2018, the average of the Top 10 Integrated Oil Majors generated a Return on Average Capital Employed (ROACE) of 11%, based on our adjusted, apples-to-apples calculation methodology. These returns are 54% correlated with the number of patents filed by each Major (chart below).
Technology leaders are implied to earn c5% higher corporate returns than those deploying industry-average technologies, which is a factor of 2x.
Upstream patent filings also predict upstream returns, with an 85% correlation coefficient. The data are skewed by one Middle East NOC, which earns exceptionally high returns on capital, but even excluding this datapoint, the correlation coefficient is 65% (chart below).
The curve is relatively flat, with the exception of two outliers, implying that it is hardest to improve general upstream returns using technology. This may be because upstream portfolios are vast, spanning many different asset-types and geographies.
Downstream patent filings predict downstream returns, with an 80% correlation coefficient (chart below). However, our sample size is smaller, as we were unable to dis-aggregate downstream ROACE for all the Majors.
The curve is very steep, indicating that downstream technology leaders can surpass c20% returns on capital, versus c10% using industry-standard technologies.
Chemical patent filings predict chemical returns, with a 57% correlation coefficient (chart below). Again, our sample size is smaller, as we could only estimate chemicals ROACEs for some of the Majors.
The curve is also steep, with technology leaders earning c10-20% returns, versus low single digit returns for less differentiated players.
Overall, the results should matter for investors in the energy sector, for capital allocation within corporates, and for weighing up the benefits of in-house R&D. We would be delighted to discuss the underlying data with you in more detail.