Decarbonisation of heavy-duty diesel engines using hydrogen fuel: a review of the potential impact on NO x emissions

As countries seek ways to meet climate change commitments, hydrogen fuel o ﬀ ers a low-carbon alternative for sectors where battery electri ﬁ cation may not be viable. Blending hydrogen with fossil fuels requires only modest technological adaptation, however since combustion is retained, nitrogen oxides (NO x ) emissions remain a potential disbene ﬁ t. We review the potential air quality impacts arising from the use of hydrogen – diesel blends in heavy-duty diesel engines, a powertrain which lends itself to hydrogen co-fuelling. Engine load is identi ﬁ ed as a key factor in ﬂ uencing NO x emissions from hydrogen – diesel combustion in heavy-duty engines, although variation in other experimental parameters across studies complicates this relationship. Combining results from peer-reviewed literature allows an estimation to be made of plausible NO x emissions from hydrogen – diesel combustion, relative to pure-diesel combustion. At 0 – 30% engine load, which encompasses the average load for mobile engine applications, NO x emissions changes were in the range (cid:1) 59 to +24% for a fuel blend with 40 e% hydrogen. However, at 50 – 100% load, which approximately corresponds to stationary engine applications, NO x emissions changes were in the range (cid:1) 28 to +107%. Exhaust gas recirculation may be able to reduce NO x emissions at very high and very low loads when hydrogen is blended with diesel, and existing exhaust aftertreatment technologies are also likely to be e ﬀ ective. Recent commercial reporting on the development of hydrogen and hydrogen – diesel dual fuel combustion in large diesel engines are also summarised. There is currently some disconnection between manufacturer reported impacts of hydrogen-fuelling on NO x emissions (always lower emissions) and the conclusions drawn from the peer reviewed literature (frequently higher emissions).


Introduction
To limit anthropogenic climate change in line with the 2015 Paris Agreement, many countries have put in place net zero greenhouse gas emissions targets.For example, the UK is now required by law to reach net zero greenhouse gas emissions by 2050. 1 To reach this, a range of low or no-carbon fuels are being considered as replacements for fossil fuels. 2 Renewable electricity is expected to be the primary source of energy in 2050, with projections indicating its generation could increase up to 500% by 2050. 3,4Battery electric powertrains are not however currently viable for some heavy-duty applications which have weight and volume restrictions, high energy density requirements and/or limits on refuelling times. 5These applications include heavy goods vehicles (HGVs), non-road mobile machinery (NRMM) and electrical generators, all of which currently rely on diesel as a practical fuel.Decarbonising these sectors requires alternatives that are competitive with diesel and reduce net greenhouse gas emissions.The Committee on Climate Change has emphasised that decarbonisation of the country's energy budget must happen quickly and at scale to reach the net zero target, 6 so solutions are required which are deliverable in the short term.
Hydrogen fuelling is a potential method for decarbonising diesel engines used in heavy-duty applications, offering similar refuelling and versatility characteristics as diesel. 7,8As an alternative to pure diesel, hydrogen can either be combusted or used in a fuel cell.Hydrogen fuel cells (HFCs) are the most environmentally benecial option from an end-use emissions perspective since their only by-product is liquid water.HFC passenger vehicles and buses are currently in use in small numbers, but more widespread uptake in buses, rail and HGVs is anticipated by 2030 9,10 as they become more cost competitive with diesel. 11Plans for the use of hydrogen to decarbonise NRMM and industrial processes are less developed.Combustion of hydrogen appears a likely short and medium-term enduse, due to simpler technological transformation requirements and lower hydrogen fuel purity requirements. 9,12,13lthough there are many signicant barriers to the development of a hydrogen economy, the UK government has recently doubled its low-carbon hydrogen production pledge to 10GW by 2030. 14It hopes for widespread use in heavy-duty transport, industry, power and heating sectors by 2050. 9Challenges such as gas pipeline conversions and the scaling of production are being considered, with investment from government, project developers and companies. 9ydrogen can be blended with diesel for combustion in existing all-diesel engines, requiring minimal structural changes to the powertrain.This allows for cheap and simple initial deployment of hydrogen and offers a low barrier to entry stimulus for the development of a hydrogen economy. 7lthough hydrogen supply will likely be limited during initial years of a (UK) hydrogen economy, the fraction of hydrogen used in a hydrogen-diesel (H2D) fuel blend could be increased as production capacity increases. 9Passenger vehicles with hydrogen internal combustion engines (H2-ICEs) seem unlikely to become a widespread reality since battery electric vehicles offer a more mature solution with cheaper running costs. 5owever, the straightforward conversion to a dual fuel engine provides a potential low-regret option in the heavy-duty sector on the path to net zero.
Unlike fuel cells, hydrogen engines are not emissionsfree.Nitrogen oxides (NO xsum of NO 2 and NO) form due to the high temperatures of combustion, as described by the Zel'dovich mechanism.The basics around mechanisms of NO x formation from H 2 combustion, air quality and emissions impacts have been reviewed recently. 13The adverse health and environmental impacts of NO 2 have an economic cost in the UK of around £5 billion a year. 15,16Application of emissions abatement and control strategies in energy and transport sectors have led to sustained NO x emissions reduction over the last 30 years, however international obligations require further NO x reductions across Europe well into the 2030s. 17Although NO x emissions from diesel engines have been under considerable scrutiny following the VW emissions scandal in 2015, 18 NO x from future hydrogen combustion has received little attention as a consequence of decarbonisation strategies. 10,19In the research literature it is common for the environmental impacts of hydrogen production to be considered in detail and on an international scale, 20,21 oen assuming that hydrogen will be exclusively used in fuel cells. 22Downstream point-of-use impacts of its combustion, especially on a national or city scale, are however less reported.
There are other lower-carbon alternatives for fuelling heavyduty engines, such as biodiesel, biogas, compressed natural gas (CNG) and liquid natural gas (LPG).Like hydrogen, these can be integrated relatively easily into existing infrastructure and much of the original powertrain can be retained.Biodiesel can be used directly in diesel engines without any modications to powertrains or petrol stations.To be truly sustainable and avoid the environmental concerns surrounding land-use change, advanced biodiesel, produced from biogenic waste and residue feedstocks, is required.The extent to which this can contribute to decarbonisation is therefore highly dependent on the amount of waste produced. 23Since waste may decline in the future due to improved management practices, 24 this energy source may only be able to contribute to decarbonisation in the short to medium term.Similarly, life cycle assessments for use of natural gas in heavy-duty engines sector suggest that deeper decarbonisation methods alongside CNG/LNG would be necessary to meet UK decarbonisation targets for 2030 and beyond. 25Hence these are only likely to be useful as interim technologies.H2D is a more intuitive option considering the potential to create the hydrogen from renewal energy (green H 2 ), and as a technology provide a bridge to later 100% hydrogen combustion or fuel-cell end use, necessary for full decarbonisation.
With any internal combustion engine, there are trade-offs between optimising for different air quality and CO 2 emissions, energy efficiency and mechanical or thermal performance. 13It has frequently been the case that efficiency and mechanical performance are prioritised, leading to some compromise on NO x emissions.If H2D or pure hydrogen is to be deployed at scale in internal combustion engines, an additional reduction in NO x emissions alongside gradual decarbonisation would be desirable as a co-benet.As an absolute minimum the expectation should be that NO x emissions should not increase if a fuel switch occurs.This is especially desirable in the UK since, with gradual electrication of the passenger vehicle eet, based on emissions inventory data, 16 non-electried diesel engines could become the major source of NO x if no further action is taken.This review presents an evaluation of the possible impacts on NO x of H2D combustion in heavy-duty diesel engines.Through qualitative and quantitative literature review, we aim to make recommendations, from a NO x emissions perspective, on the optimal deployment of hydrogen for decarbonisation of NRMM, HGVs and power generators.For completeness, an evaluation of changes in other air quality relevant pollutants are also presented.

NO x formation control
In high temperature combustion applications, NO accounts for around 95% of total NO x emissions at point of exhaust. 26During the combustion of diesel in Compression Ignition (CI) engines, most NO x forms through the Zel'dovich mechanism: 27 In a CI engine, this mechanism is elevated by high temperatures, high oxygen concentrations and extended residence time of nitrogen in high temperature regions of the combustion chamber. 28In theory, adding hydrogen to the combustion process has two opposing effects on NO x formation.The higher adiabatic ame temperature of hydrogen acts to increase NO x emissions through the Zel'dovich mechanism.However, the higher ame velocity means the high temperature conditions exist for a shorter period of time, reducing NO formation.If hydrogen is added through the intake air manifold, it can reduce the amount of oxygen entering the chamber, also reducing the yield from the Zel'dovich mechanism.The overall effect of hydrogen addition on NO x emissions depends on these competing processes.The balance of effects will depend upon many operational factors such as engine load, speed, combustion-system design and fuel injection parameters. 29For pure diesel combustion, NO x increases with load due to elevated combustion temperatures, but decreases with engine speed due to reduced residence time of combustion gases. 30O x emissions are also inuenced by relative injection timings of diesel and hydrogen.NO x is produced in high temperature zones of the combustion chamber, whose existence depends on the heat release rate.NO x formation can be limited by increasing the ignition delay (the time between start of fuel injection and start of combustion). 31Further complications therefore arise if the presence of hydrogen affects that ignition delay time.
The emissions of NO x from diesel combustion can be abated to a degree by internal measures combined with exhaust aertreatment, techniques.Both approaches can also be applied to H2D engines.Aer-treatment such as selective catalytic reduction (SCR) and lean NO x traps (LNT) are efficient methods that can reduce NO x in dual-fuel applications, 13 although they increase cost and complexity of the combustion system and add to overall equipment costs. 32Internal measures include fuel lean conditions, water injection and exhaust gas recirculation (EGR).Although these methods act to produce a cleaner combustion system, each has its trade-offs.Fuel-lean conditions inherently reduce engine efficiency; 13 water injection causes elevated CO and unburnt hydrocarbon emissions; 33 and high EGR rate causes elevated emissions of particulate matter (PM). 34drogen fuelling in small engines Initial development of hydrogen combustion engines focused on use in light-duty passenger cars with spark ignition (SI) engines, 35 pioneered particularly in Japan.Although NO x emissions from small SI engines are lower than their gasoline equivalents, mechanical performance is signicantly limited at high loads due to engine knock and autoignition problems. 29,36ese factors make SI engines generically unsuitable for heavyduty applications, and so our later analysis considers CI engines only.More broadly, smaller SI hydrogen engines lack substantial competitive advantages over battery electric vehicles, which now dominate sales of low-carbon passenger cars.
There is however considerable literature on air pollution emissions from small CI engines running on H2D dual fuel, both single-cylinder test-engines and those found in passenger cars.This can provide helpful insight for interpreting possible outcomes from H2D use in larger (CI) engines, for which the literature is noticeably less extensive.Dimitriou and Tsujimura reviewed the performance and emissions of CI engines run on H2D dual fuel up to 2017. 37Almost all experiments considering NO x emissions were based on either a low-power single-cylinder engine or a multicylinder passenger car engine.Experiments varied in engine design, load, speed, fuel injection method and hydrogen fraction range.There was general consensus that hydrogen addition benecially decreased emissions of CO, CO 2 , PM and SO 2 , however results for NO x were more mixed.Compared to diesel-only operation, studies could be found that reported decreases in NO x emissions, increases in emissions or little change.There also appeared to be no clear correlation between EGR rate and NO x emissions. 34A similar conclusion was drawn in another review, 38 which suggested the highly variable results were due to substantial differences between test facilities and accuracy of simulation methods.
Engine loading does however appear to be a key factor that in past literature has affected NO x emissions in single cylinder test-engines.This is likely to translate to larger multicylinder engines due to its basic dependence on combustion chamber temperature.However, whilst some studies report that hydrogen addition caused an increase in NO x emissions at high loads, 39,40 others found a reduction, even without an anticipated reduction in efficiency. 41This hints at the complexity of the dual-fuel combustion process.The most common hydrogen injection applied in dual-fuel CI engine studies have been port fuel injection (PFI), manifold injection and direct injection (DI).PFI and manifold injection methods are associated with combustion limitations and reduced volumetric efficiencies. 42I is a more recent concept aimed at overcoming these performance limitations.In an SI engine, DI fueling has also been shown to produce lower NO x emissions when compared to PFI methods under high load conditions. 43However, the use of DI requires more modications to the original diesel engine. 44A recent CI engine simulation suggested that NO x formation only falls lower than a diesel-only case with a hydrogen energy share above 80%, 45 arising due to improved fuel mixing and the increase in ignition delay caused by hydrogen addition.

Hydrogen fuelling in large engines
Prior to this review there were no assessments of the potential air quality effects of hydrogen-fuelled heavy-duty diesel engines, a key gap in evidence given the current need to develop both policy and regulation in this area for net zero.The available literature reporting NO x emissions from large diesel engines (typically used in NRMM, HGVs and generators) when run on  H2D dual fuels are summarised in Table 1.These works are less numerous than those studying NO x from passenger car engines, but importantly are oen more recent.Most studies involve heavy-duty road vehicles, with very few papers specically addressing NRMM or industrial engine/prime power.][48][49][50][51] Although NO x is the primary pollutant of interest in our review, it is important to cnsider other point-of-use emissions arising from H2D.In all cases CO 2 emissions decrease linearly with hydrogen addition by energy share, due to the reduction in the amount of carbon-containing fuel.In most cases, CO decreased for H2D combustion compared to diesel due to increased H/C ratio, improved air-fuel mixing and shorter combustion duration.However, the studies corresponding to datasets 2 and 13 found CO emissions increased on hydrogen addition, due to the increased ignition delay reducing temperatures enough to increase the extent of incomplete combustion. 38,50Reduced oxygen content also contributed to this through a reduction in CO oxidation. 50Studying Table 1, this is likely to be due to engines being run at full load, causing increased total fuel consumption.
PM emissions generally decreased for H2D compared to diesel, due to a reduced diesel-to-air equivalence ratio.The reduction was usually most signicant at low loads for this same reason.Exceptions were commonly due to decreased soot oxidation rate and incomplete combustion, especially at low loads. 49,52,53esults for hydrocarbon emissions are somewhat discordant and appear to be dependent on operational parameters affecting the amount of diesel being burned and the owrate of diesel.Decreases were observed for studies which were accompanied by increased combustion efficiency, these decreases commonly being largest at high hydrogen energy shares and low loads.An overall increase in HC emissions was observed when the extent of incomplete combustion increased. 38,49,52,53Experimental variation, including the presence/absence of EGR, is likely to have inuenced results here. 46imilar to results seen in smaller engines, the relationship between NO x emissions and hydrogen fraction is very variable in literature reports, likely impacted by experimental factors such as engine model, hydrogen injection method, load, speed and the application of exhaust gas recirculation.5][56][57] The observed variation in NO x emissions from hydrogen addition compared to pure diesel combustion were small, and there was oen no clear trend.Other studies have showed that the trend in change in NO x emissions was only apparent for higher hydrogen fractions. 53[51][52][53] Most found that when hydrogen is added to diesel, NO x emissions tend to decrease at low loads and increase at high loads.However, different experimental conditions mean that the denition of what constitutes 'low' and 'high' load varies between studies, as well as the   53 Zhou et al. also found that ignition delay varies with both hydrogen addition and load, affecting NO x through altering heat release characteristics. 49An increased ignition delay at low load would result in a narrower heat release peak, resulting in reduced thermal NO x formation. 58Kumar et al. used supporting temperature, oxygen and unburned hydrogen measurements to explain the load-dependency of the effect of hydrogen on the combustion process. 51Lower temperatures, reduced oxygen content and increased unburnt hydrogen emissions suggested that at low loads, hydrogen acted as a heat sink due to its higher specic heat capacity.At higher loads, reduced emissions suggested it acted as a heat source to enhance the combustion process.We suggest that at higher load, the effect of load on temperature becomes more important than the effects of hydrogen addition which cause a reduction in temperature, thereby causing an overall increase in thermal NO x formation.
A similar effect of increased ignition delay was also observed by Hosseini and Ahmadi, who conducted a numerical investigation of DI of hydrogen into a heavy-duty engine. 38A large and approximately linear decrease in NO x emissions was observed when hydrogen energy share was varied from 0 to 70% at full load.Similar results were found for both hydrogen addition and hydrogen replacement cases, with larger emissions reduction found in the replacement case due to a larger reduction in temperature.DI was found to increase ignition delay signicantly, which may explain the observed decrease in NO x emissions.Only results from the substitution case are provided in Table 1 because addition of hydrogen does not result in decreased diesel usage, something with no benets for reducing carbon emissions, and hence outside a net zero scope.
5][56] The expected increase in NO x emissions with load was observed at small hydrogen energy shares of up to 2%, but not for higher hydrogen intake.Inferring trends has likely been complicated through application of EGR and the fact that only small hydrogen energy shares of up to 5% were tested.Other studies also found that the application of EGR complicated results.For example, Liew et al. compared NO x emissions from two heavy duty diesel engines, one with and one without EGR. 52,53For the engine without EGR, NO x decreased at low loads and increased at high loads as hydrogen intake increased.However, this effect was not observed for the engine with EGR, and unexpected changes in EGR ow rate were observed at times.Hydrogen inclusion changes exhaust gas composition, making it difficult to predict how temperature, and therefore NO x emissions, are affected by EGR when hydrogen is present.
Only two papers have reported the effect of engine speed on NO x as hydrogen fraction was added. 50,57NO x emissions were higher at the lower engine speed, as expected.Increasing hydrogen intake slightly reduced NO x at low engine speeds, whilst dramatic increases in NO x emissions were observed at high speeds.The authors acknowledged that the impact of hydrogen on NO x emissions varied depending on engine speed, but did not propose an explanation for this.The complexity of dual fuel combustion is evident and current understanding required for accurate emissions prediction is not complete.Despite this, some important conclusions can be drawn: NO x emissions tend to decrease at low loads and increase at high loads when the hydrogen fuel fraction is increased.
The presence and rate of EGR complicates the relationship between hydrogen fraction and NO x .
If results are to be useful for future policies supporting H2D combustion, experimental conditions need to mimic real-world diesel engine operation.This could give insight into the most suitable areas for H2D diesel engine combustion from a NO x emissions perspective, since the load at which a large diesel engine will run depends on its end-use application.For example, electrical generators and other stationary machinery tend to run at relatively constant, higher loads, typically above 50%. 59[61]

Quantitative analysis
We conduct a meta-analysis of suitable datasets from Table 1.The amount of hydrogen added is expressed in the literature either as an energy share percentage (e%) or as a fractional volume in air (vol%).Hydrogen e% was chosen as the common axis, as it takes into account the difference in energy densities of hydrogen and diesel and meant that most of the literature could be used.Where the unit used in the original paper was vol%, conversions to e% were completed where possible, using a provided calibration curve.Datasets 6 and 13 were excluded on this basis.Up to 40 e% hydrogen addition was considered.Literature suggests this amount of hydrogen can be added safely and easily to an existing diesel engine at a range of loads. 53,62,63In addition, it is unlikely there would be sufficient industrial hydrogen production capacity in initial years for more than 40 e% hydrogen to be used widely on a national scale.Datasets 7-11 were excluded because they only provide data for small e% well below 40 e%.The literature reports NO x emissions in a range of different units.In most cases, this was not an issue, since NO x was converted to a change in emissions relative to pure diesel combustion in the same engine.NO x emissions in dataset 3 required conversion from a ratio of H2D emissions to pure diesel emissions, to a fractional change.
Least squares regression analysis was performed on each dataset to give a simple expression of change in NO x for different hydrogen fractions.Only 0-40 e% points were included in the regression analysis because the literature review revealed that the effect of hydrogen addition on NO x emissions can change at very high loads (e.g.dataset 5).A linear relationship under these hydrogen fractions, whilst not wholly accurate, is suitable in providing a range of possible outcomes for NO x , especially when combining multiple datasets.
The data was split into 0-30% load and 50-100% load, to approximately correspond to the use characteristics found for mobile machinery (0-30) and stationary engine (50-100) end uses.For datasets 1 and 14, where loads were not provided, the low load cases were assumed to be within the 0-30% range and the high load cases in the 50-100% range.This assumption was considered preferable to excluding the data, due to the limited number of datasets.Dataset 14b was a medium load case and therefore excluded from the analysis.Best-case, median and worst-case changes in NO x emissions were calculated for hydrogen fractions of 10, 20, 30, 40 e%.Best-case is the largest reported literature reduction in NO x emissions compared to diesel combustion, whilst worst-case refers to the largest increase.Focusing analysis on these three cases reduced the error associated with those few datasets whose regression analyses produced low R 2 values.

Results and discussion
Fig. 1a and b show the linearised responses from literature of NO x emissions from H2D combustion in large diesel engines for low (0-30%) and high (50-100%) load, respectively.There is considerable variation in both gures, which increases with hydrogen energy share, due to differences in experimental factors across studies.For 40 e% hydrogen, NO x relative to diesel combustion varies over the range À59% to +24% for the low load case and À28 to +107% for higher load applications.Fig. 1c and d show best, worst and median scenarios for NO x emissions for 4 different hydrogen fractions.Whilst this analysis does not provide information on what is the most likely scenario, it does guide the level of risk and potential benets/ dis-benets, for NO x emissions that are associated with different H2D applications.
An initial addition of 10 e% hydrogen for low load applications appears to be the lowest-regret option.Fig. 1c suggests NO x emissions would only increase by 6% in the worst-case scenario.If only initially deployed in industrial NRMM, this would increase total UK NO x emissions from fuel combustion activities by $0.2%. 16This is a negligible change placed in the wider context of NO x emissions decreasing by about 3% per year in recent years. 16Aside from the reality that blending hydrogen initially in small amounts is likely an economic necessity to facilitate a later full transition to hydrogen, a potential upside reduction of NO x up to 15% makes this a reasonable rst step.
Fig. 1d shows the same hydrogen addition for high load applications comes with higher risk in terms of NO x , with a possible 27% increase in emissions in the worst-case scenario for only a 10 e% fuel share.This is higher than the worst-case Fig. 1 Meta-analysis of effect on NO x emissions for H2D combustion in large diesel engines for different H2D fuel compositions.(a) 0-30% low load case (b) 50-100% high load case.Numbers in the legend correspond to datasets in Table 1.The median dataset is presented as a red dashed line.(c) Worst (grey), median (orange) and best-case (blue) NO x scenarios for 0-30% load (d) worst (grey), mean (orange) and best-case (blue) scenarios for 50-100% load.
scenario of 40 e% hydrogen at low load.The potential reward in emissions reduction is also much smaller, at just 7.5%.This analysis suggests that H2D combustion in large diesel engines would be best used in applications which have lower average loading.This would include a range of NRMM typically used on construction sites, such as excavators and dumpers.Hydrogen energy shares should be low at rst, until NO x emission factors could be evaluated in the practice in the eld.Using H2D in electrical generators, which operate at higher loads for longer periods of time, appears more likely to lead to increase in NO x when compared to current pure diesel engine emissions.
Whilst there is a clear difference in results for high and low load cases, the limited database should be kept in mind.Additionally, we have linked load to application by reported average engine load.In reality, especially for mobile engine applications, the engine will run at a wide range of loads depending on driving conditions/power demands.
The effect of age/era of the test engine as a contributing factor to the wide range of NO x emissions reported has been considered.Newer engines are designed to be compliant with more stringent NO x regulations, hence older engines in the analysis are likely to have produced more NO x because of their design.Although the above meta-analysis uses relative NO x emissions, it is likely the effects of hydrogen fraction change depending on engine era due to different engine designs.Since production year is not available for most of the engines in the literature, it is not clear how important this consideration is.All literature used has been published in the last 13 years.NRMM tend to have long working lifetimes, so it is plausible that some older engines might be retrotted to accommodate hydrogen, particularly large stationary installations such as diesel farms.

Hydrogen fuel and engine idling
Engine idling is oen characterised by elevated exhaust emissions, including for NO x . 47,59,62Dataset 6 (Table 1) is the only study to explicitly consider NO x emissions with hydrogen addition under idling conditions.A 7% decrease in NO x emissions was reported as hydrogen fraction was increased to 1.2 vol%.The only other literature was based on a marine diesel engine, with a similar effect reported. 63Results from a two-year study of NRMM in real-world operating conditions on a construction site suggested that, on average, 45% of plant time is spent idling. 60This is much higher than for road transport and a lack of regulations to reduce idling of NRMM make air quality emissions a serious consideration in cities. 64 Desouza et al. found exhaust treatment technology ineffective during idling 60 since exhaust systems oen rely on heat from the engine to maintain catalyst temperature.The data from Jhang et al. suggested adding hydrogen could be an alternative method to reducing NO x emissions under idling conditions.Whilst more data at higher hydrogen fractions and from different engines would be necessary to conrm this, there is no evidence suggesting hydrogen addition would increase idle emissions.This potential reduction in idling NO x emissions from hydrogen addition further supports the case for H2D use targeted at low average load applications.

Exhaust gas recirculation
EGR is a method commonly used in modern diesel engines to reduce NO x emissions by reducing combustion temperatures and the oxygen content of intake air. 65It has the potential to reduce NO x from H2D combustion to facilitate its use in high load applications such as generators.The studies by Liew et al. on NO x from H2D provide the best quantitative comparison of diesel engines with and without EGR, since other experimental conditions are consistent across studies.Fig. 2 presents analysis of datasets 4 and 5, of relative NO x from H2D containing 40 e% hydrogen compared to pure diesel combustion.There appears to be no clear, predictable trend for how relative NO x is inuenced by EGR at different loads.Relative NO x is considerably lower when EGR is present at 10 and 70% loads, but this is not the case for intermediate loads.At 15 and 20% loads, relative NO x emission more than doubled and EGR had no effect on relative NO x at 50% load.Both the presence of hydrogen and engine load alter exhaust gas composition, which is likely to be a factor.
Whilst the data at 70% load suggest EGR could reduce NO x , and thus support a case for H2D use in high load applications, two studies is a very limited evidence base.In addition, studies have found that for engines running on H2D, EGR reduces engine efficiency and increases PM emissions. 37,46H2D in low average load applications is still recommended as the low-regret option of hydrogen combustion in large diesel engines.Fig. 2 suggests that EGR may not be necessary in these cases and may in fact signicantly increase NO x emissions.Again, the limited data available means this is not denitive.

Commercial review
When and how decarbonisation of large diesel engines will be achieved relies largely on industry research and development of new technologies and alternative fuels.The main commercial organisations considering large diesel engine decarbonisation Fig. 2 NO x emissions from combustion of H2D dual fuel of composition 40 e% hydrogen, for engine loads 10-40%.Raw data for the engine with EGR (blue) is taken from dataset 4 (see Table 1).Raw data for the engine without EGR (orange) is taken from dataset 5 (see Table 1).

JCB and Cummins
In the UK, JCB has been a leading proponent of decarbonisation of construction machinery.Cummins, an American engine manufacturer, have been developing heavy-duty H2-ICEs since 2021.Both companies produce a range of small electric machines, such as forklis, dumpsters, and small trucks. 66,78owever, they identify hydrogen as the solution for larger NRMM and long-haul trucks, which require higher loads and long work times.Their H2-ICE approach is to produce new engines to run on 100% hydrogen, rather than converting current engines for dual fuel use.Whilst this means a completely new engine is required, it utilises existing engine know-how and production facilities, and many of the powertrain components can be retained. 72,74Like other manufacturers, 95,96 both companies report exploring HFCs for large vehicles. 67,75For heavy-duty applications, the focus on H2-ICEs has been more recent, with manufacturers identifying it as the most cost-effective method for heavy-duty engine decarbonisation. 70,73al fuel combustion Several companies are now offering diesel engine retrots for H2D dual fuel operation.Retrots are generally advertised for heavy-duty road vehicles, but some companies suggest the technology can be applied to other applications. 88Hydrogen storage tanks are integrated such that space within the vehicle is generally not reduced and the whole retrot adds modestly to the vehicle weight. 82,88Whilst yielding an initial reduction in CO 2 , these retrotted engines cannot support a nal 100% transition to hydrogen.However, companies are aiming to increase the amount of hydrogen their retrots can support. 82,92 downside to these retrots are the lead times, with one company suggesting retrots to a new engine can take up to 6 months. 88ydra, a Canadian company, introduced hydrogen to the engine by manifold injection, 82 similar to most of the research literature in Table 1.They claim there is no loss in engine performance, whilst tailpipe CO 2 emissions from each vehicle are reduced by up to 40%.Both the Canadian government and Hydra have put in place incentives to encourage users to transition to dual fuel. 82,97Whether government incentives are best targeted at HGVs or NRMM may depend in part on the relative contribution each sector makes to emissions: HGVs contribute a higher fraction of GHG emissions from transport in Canada (38% in Canada, compared to 27% in UK) due to greater mileage. 98,99oth HYDI and HyTech Power use on-board electrolysers supplied with water to produce hydrogen which is injected into the air-fuel mixture prior to combustion. 85,92][102] HYDI have reported an increase in power and improved fuel efficiency. 84The system takes 4-8 hours to t and does not require engine modication, since only a small energy share of hydrogen is used. 86

Exhaust emissions
Performance for NO x and PM emissions are commonly recognised as key issues by manufacturers considering hydrogen fuelling.Many claim a reduction in emissions compared to pure diesel combustion, from either real-world or in-lab tests. 87Achieving lower NO x emissions using aer-treatment technologies rather than altering engine design, is reported widely. 80,93JCB reported a 98% reduction in engine-out NO x emissions and no reduction in performance for hydrogen fuel, as compared against a pure diesel comparison with aer-treatment technology system. 70It is therefore possible that latest engineering innovation has overcome some of the compromises between engine performance and tailpipe emissions.However, there is clearly a disconnection between this level of manufacturer proposed reductions of NO x emissions (in Table 2) and the conclusions that would be drawn from the peer reviewed literature (earlier gures).
For dual fuel engines that run on a range of fuel compositions, the amount of hydrogen can be varied.For example, New Holland dual fuel tractors increase the hydrogen fraction at lower loads. 94Results from the meta-analysis in this paper indicated that, when considering NO x emissions, lower loads are indeed more suited to dual fuels and higher hydrogen fractions.Since technology exists to control the amount of hydrogen added based on engine load, and this could make dual fuel suitable for a range of applications.This may inevitably add cost however to combustion retrots, which would need to be weighed up against the benets from NO x emissions reduction.

Conclusions and recommendations
Hydrogen and H2D dual fuel combustion are potential solutions for reducing greenhouse gases from diesel engines when used in heavy-duty diesel applications.H2D dual fuel may become an important short-term decarbonisation method as part of a longer-term transition to pure hydrogen technologies.Such an approach offers a relatively cost-effective way to reduce fossil fuel reliance as it can largely be used in current diesel engines, with minimal modication.Particular attention in this review is paid to NO x from H2D combustion in large CI engines, an air quality impact which does not arise from alternative powertrains such as fuel cells and batteries.
Peer-reviewed studies of large CI engines run on H2D are limited.The relationship between NO x emissions and hydrogen fraction can difficult to discern due to experimental variation across studies.It was clear, however, that engine load and the application of EGR were key factors in affecting results.A range of possible outcomes for NO x were found, resulting from H2D combustion of up to 40 e% hydrogen in large CI engines.A meta-analysis was used to quantify the range of possible NO x emissions resulting from H2D combustion, of up to 40 e% hydrogen, in large CI engines.A range of possible outcomes for NO x were found, from a small decrease to a large increase.The median increase in NO x emissions from H2D (compared to diesel) was smaller for low loads than high loads.Considering the highest reported NO x emissions, H2D of 10 e% hydrogen under high load was worse than 40 e% hydrogen under low load.A quantitative comparison of similar engines with and without EGR revealed no clear trend for NO x under different loads.It is possible that EGR can reduce NO x at both very high and very low loads, but there is not enough data to conrm this.Without additional abatement/aertreatment technologies, H2D would be best used in lower average load applications such as excavators, dumpsters and cranes.It should be noted that this conclusion is based on a relatively limited database, hence we suggest that more research and testing should be conducted to conrm this result.
EGR may be effective for reducing NO x at both very high and very low loads when H 2 is added, but there is not enough data to have high condence in this conclusion.The impact of EGR is likely to vary depending on the engine application, hence choices around the use of H2D in a hydrogen economy should be matched to specic end-uses.Similarly, the use of higher hydrogen fuel fractions at idle may help reduce NO x from construction machinery, a notable benet given these oen occur in urban areas.
More data is needed on the effects of other aspects of engine operation on NO x emissions, such as engine speed and hydrogen injection method.This may help to separate individual NO x contributions to gain a better understanding of the relationship between engine operation, fuel composition and NO x emissions.Additionally, conrmation through experiment is needed to establish that engine load is the dominant factor controlling NO x emissions from H2D combustion in CI engines.
A follow on from this review would be the creation NO x emission scenarios from the meta-analysis with studies weighted according to the different engine designs likely to be used in the UK (or any country looking to use H2D in heavy duty engines).4][105] A larger range of studies would be needed for this analysis if uncertainties in projections were to be narrowed.
Both H2-ICE and H2D dual fuel retrots are being demonstrated commercially, many of whom are also developing green hydrogen production and supply routes.NO x emissions are explicitly considered by manufacturers and are claimed to be reduced either by internal or external measures.For example, one dual fuel manufacturer used adaptive technology to reduce hydrogen fraction as engine load increased.It was not clear from public information sources whether this was to specically limit NO x emissions, however it may well have had this bene-cial effect.This showed that if the cost is reasonable, additional technology could be applied to actively control hydrogen fraction based on engine load, and to minimise NO x emissions.
Taking the evidence in this review the following ve recommendations for policy consideration emerge: 1. H2D as a technology would be best used in lower average load applications such as excavators, dumpsters and cranes if avoidance of NO x emissions was a major consideration.The scale of NO x benets is difficult to judge and would depend on the sophistication of the aertreatment system supplied.Replacement of diesel with H2D would however likely provide more universal reductions in emissions of SO 2 and PM including in high load applications.
2. Technology could be applied to actively vary the hydrogen fraction used based on engine load, such that NO x emissions were minimised.Use of higher H 2 fuel fractions during periods of idle may help reduce NO x emissions, particularly in construction applications.This would be a notable benet given these engines are oen used in urban areas.
3. There is uncertainty about the impacts of EGR regimes and their application in H2D dual fuel combustion engines.The impact of EGR is likely to vary depending on the engine application, and research is needed to understand how best to match EGR with each end-use.
4. More evidence is needed to determine whether hydrogen addition to diesel would benet NO x emissions under idle conditions for a wider range of engine applications (beyond the construction sector) and if used in combination with abatement/aertreatment approaches.
5. The long-term trajectory for the use of hydrogen as a fuel may be inuenced by an early adoption of H2D dual fuel in sectors such as capacity market power generation, construction and agricultural sectors.Moving later from H2D to H2-ICE, rather than fuel cell power trains, would be an incremental progression that would allow manufacturers to continue to exploit past investments in ICE production facilities and existing technological know-how.Policy support for early adoption of H2D may set a pathway that retains combustion appliances for the longer-term, along with the need to manage their possible NO x air quality impacts.
which NO x emissions change.At low loads, Liew et al. observed a delay to the start of combustion on hydrogen addition, possibly explaining the NO x decrease observed.

Table 1
Summary of test conditions and pollutant emissions from literature investigating hydrogen and diesel dual fuel combustion in heavy-duty compression ignition engines

Table 2 .
Since these are business developments, the basic commercial intentions are generally issued via investor announcements or press releases, but rarely are any detailed datasets provided.None of the manufacturers in Table2has provided data on emissions in open-source formats, or via the peer-reviewed literature.