October 2015

Heat Transfer

NOx, NOx, who’s there—steam cracker or SMR?

This article addresses the generation of thermal NOx in the furnace of each process, NOx abatement by SCR, and premature degradation of SCR catalyst by chromium species in the flue gas.

Kunz, R. G., RGK Environmental Consulting, LLC

The remaining pieces of the puzzle have fallen into place for the startup and operation of selective catalytic reduction (SCR) in an ethylene plant in the US. It is now instructive to revisit nitrogen oxide (NOx) issues in steam cracking and steam methane reforming (SMR) with updated information.

NOx from combustion in both processes is strikingly similar, such that experience gained in one readily transfers to the other. This experience transfer is especially important because more stringent NOx limits for many types of sources are anticipated.

This article addresses the generation of thermal NOx in the furnace of each process, NOx abatement by SCR, and premature degradation of SCR catalyst by chromium species in the flue gas.

Ethylene manufacturing process description

Thermal cracking of hydrocarbons in the presence of steam is the primary commercial route to ethylene and its coproducts.1 The ethylene cracking unit is also called a steam cracker, an ethylene cracker, a thermal cracker or a pyrolysis furnace. A wide variety of feedstocks ranging from ethane to gasoil can be used. Reaction occurs inside empty tubes, known as radiant coils, suspended in a fuel-fired furnace.

In addition to ethylene, cracking units generate other products, with greater amounts produced from the heavier feeds. At the same time, unwanted coke is laid down on the inside surface of the coils and interferes with the operation. As a result of coke formation, elements of the process train must be taken out of service periodically and decoked using a mixture of steam and air.

Steam cracking produces an impure methane fuel, known as methane-rich gas, or the pyrolysis methane fraction, plus an impure hydrogen-rich gas, which may be upgraded for chemical uses or also used as fuel. This fuel may contain byproduct ethane and propane.

Cracking of heavier feedstocks results in appreciable quantities of liquid products. These include pyrolysis gasoline (pygas), a gasoline-like liquid high in unsaturated compounds and rich in aromatics, along with a heavier cut called pyrolysis fuel oil. Both of these liquids are unstable in storage and incompatible when blending with refined products from other sources. Pygas has the potential for a chemical feedstock upgrading. Some ethylene furnaces are capable of burning pyrolysis fuel oil.2

An ethylene cracking furnace can also use other fuels available in a petrochemical complex. Natural gas, refinery fuel gas (RFG) and distillate fuel oil are the most likely fuels.

SMR process description

SMR reacts a desulfurized hydrocarbon feed with steam to produce synthesis gas (syngas), a mixture of hydrogen, CO and CO2. Reaction occurs at elevated temperatures over a solid nickel-based catalyst contained inside tubes suspended in a furnace.1 Major process steps consist of feed sulfur removal, reforming, water-gas shift and product purification. Feed is usually natural gas, but it can also be refinery gas, propane, LPG, butane or straight-run naphtha. Additional steam is provided downstream in one or more shift-converter vessels outside the furnace to maximize the yield in an H2 plant. In an H2-CO plant, coproduct CO is recovered without shift by low-pressure cryogenic distillation following CO2 removal by regenerative amine absorption and a drying step.

In either case, H2 product is separated from the syngas in a pressure swing adsorption (PSA) unit capable of producing a purity of 99.9% to 99.999%. Other components in the syngas end up in the purge gas resulting from the periodic regeneration of the PSA unit. H2 plant purge gas contains unrecovered H2, unreacted methane, excess steam, CO, CO2 and impurities such as N2 from the feed. Combustible components include H2, CO and CH4. Purge gas containing approximately 40% CO2 recycles to the reformer furnace as the bulk of its fuel. Typically, natural gas or RFG3 supplements the purge gas. In contrast, hydrogen-CO plant purge gas/furnace fuel contains approximately 90% H2 and 10% CO without high CO2.4

Older H2 plants, built before the mid-1970s, employed amine absorption, CO2 removal and methanation to make H2 of lower purity, along with high-purity CO2, either liquefied for sale or used in further processing. Such plants are fired solely on external fuels. Table 1 summarizes fuels for the two different processes, while Table 2 shows where combustion is similar.



NOx correlation for SMR furnace burners

Previously published SMR NOx data and plant-specific correlating curves are shown in Figs. 1–3.3–5 These data are the basis for a generalized thermal-NOx correlation featuring a functional form based on kinetic theory, with empirical constants derived from regressing numerous field data.

  Fig. 1. SMR furnace NOx: H2 purge gas 
  blends, conventional burners,  and ambient 
  air3 (AFT and NOx are  less vs. natural gas).


  Fig. 2. SMR furnace NOx: H2-CO purge gas
  blends, low-NOx burners, and ambient air4
  (AFT and NOx are greater vs. natural gas,
  and even higher with added H2).

  Fig. 3. NOx from natural gas-fired  SMR
  furnaces, burners and air  preheat, as noted.3–5

The author first presented this original correlation at the 1992 NPRA (now AFPM) Annual Meeting, and later summarized it in this publication.3 The availability of more data enabled further correlation elaboration. The resulting equations are plotted in Fig. 4.4 Adiabatic flame temperature (AFT) and furnace excess O2 serve as surrogates for the actual flame temperature and O2 concentration in the flame, where the thermal NOx reactions take place.

  Fig. 4. SMR NOx correlation.4

The theoretical adiabatic flame temperature is that calculated when a fuel is burned, without mechanical work or gain or loss of heat, to the theoretical end products, regardless of any equilibrium condition that might apply. It depends on the fuel’s heating value, the combustion products, and combustion air and fuel temperatures.

Fig. 4 contains two virtually parallel lines for conventional and low-NOx burners only, dictated by data availability at the time. NOx is in ppmv, and O2 is in vol%. The basis is either both wet or both dry.

Extension to ethylene cracking furnaces

By considering differences in fuels and conditions, the SMR NOx correlation can be modified to apply to ethylene furnace burners (Fig. 5).6–8 AFTs for gaseous ethylene furnace fuels lie at the high end of the range for SMR fuels. (Decoking is not modeled.)

  Fig. 5. Proposed NOx correlation for
  ethylene furnaces.

Ethylene furnaces also run hotter (Table 3),1 averaging 2,000°F (approximately 1,100°C) and 1,850°F (approximately 1,010°C), respectively. Therefore, at the same AFT, about 1.5 times as much NOx would be generated from the typical ethylene furnace compared to the typical SMR furnace. The actual furnace temperature has been used by others as an additional parameter to improve NOx predictions.


Experimental verification of ethylene NOx predictions

General, historical or anecdotal NOx information is easy to find. However, open-literature full-scale ethylene-furnace NOx data accompanied by simultaneous operating conditions are few and far between. NOx testing in a burner manufacturer’s pilot facility often produces low estimates compared with full-scale operation, and tests of multiple burners have produced significantly higher NOx levels than single-burner tests in a smaller furnace.7

Still, it is possible to identify some usable cases for comparison,6–7 with results summarized in Fig. 6. Observed NOx runs the gamut from approximately 100 ppm for conventional burners, to ultra-low-NOx burners at 28 ppmd–32 ppmd for low air humidity and 9% H2 in the fuel.9 To be included in the parity plot, data must originate from commercial ethylene furnaces with conditions spelled out well enough to validate the model unequivocally. Agreement is satisfactory, but caution should be exercised because of limited data.

 Fig. 6. Parity plot—predicted vs. observed NOx.

Proposed correlation for SMR ultra-low-NOx burners

Ideally, one would regress a body of commercial SMR ultra-low-NOx data to generate an additional line in Fig. 4. Absent access to such data, a different approach is necessary. Fortunately, the ethylene NOx correlation can be used to “double back” and estimate a relationship for SMR ultra-low-NOx burners. The ethylene furnace equations suggest an NOx ratio of approximately 3:1 between conventional and ultra-low-NOx burners, resulting in the proposed correlation shown in Eq. 1, also to be used with caution:

ln (NOx / O2) = 11.5 – 3.6 × 10,000 ÷ AFT(°R)     (1)

Predicted NOx was compared with data reported for the replacement of conventional burners with ultra-low-NOx burners in a commercial SMR furnace.10 NOx predictions of 24 ppmd–28 ppmd (conventional) and 8 ppmd–9 ppmd (ultra-low-NOx) for an assumed range of 1.5%–3% excess O211 agree well with the 26 ppmd and 10 ppmd reported.

Selective catalytic reduction (SCR)

When burners alone cannot meet regulatory requirements, SCR is used. The lowest NOx found from ultra-low-NOx burners in ethylene service is on the order of 30 ppmd vs. 10 ppm for SMR.

In the SCR process, the NO and NO2 making up NOx react with ammonia (NH3) to form N2 and water vapor in the narrow passages of a flow-through catalyst.1, 12 A typical temperature range for base metal catalysts is 600°F–750°F (316°C–399°C), as shown in Eqs. 2 and 3:

4 NO + 4 NH3 + O2 “ 4 N2 + 6 H2O    (2)

NO + NO2 + 2 NH3 “ 2 N2 + 3 H2O     (3)

Injected ammonia and the NOx components adsorb and react on the inside surface and in the pores of the catalyst. Reaction products then diffuse back into the flue gas stream to emit from the stack. Reaction stoichiometry depends on the relative amount of each oxide and whether O2 is present. In the absence of competing side reactions, the theoretical molar ratio of NH3 reacted to NOx destroyed is 1.

Excess ammonia added appears in the effluent and is designated as ammonia slip. Sufficient catalyst must be present to provide the required NOx removal at an acceptable level of slip. Removal efficiency depends on the amount of catalyst; the NH3-to-NOx ratio; and the local distribution of ammonia, NOx and flow across the SCR inlet.

SCR experience on SMR furnaces by one industrial gas supplier includes six plants in California and Texas.12–14 Other H2 manufacturers have also installed SCR on the reformer furnace.15, 16 SCR is proven in ethylene plants—two in Japan2, 17 and the latest in Texas.15, 16–19 Outlet NOx and ammonia slip concentrations of 10 ppm or less have been demonstrated on both SMR16 and ethylene19 furnaces.

Degradation of SCR catalyst by chromium species

A loss in catalyst performance coupled with a buildup of chromium species on the SCR catalyst in both an SMR plant12 and steam crackers2, 17 was noted approximately 10–15 years ago. The activity of the initial charge of catalyst deteriorated faster than expected.

A comprehensive investigation to explain this phenomenon in SMR furnace flue gas was subsequently presented and published.1, 13, 21 In that study, it was speculated that the poisoning chromium species in SMR service were being similarly generated on the exterior of ethylene furnace coils exposed to hot flue gas. Both kinds of plants use the same family of chromium-nickel alloys,1 and tube/coil metal temperatures, although different, are not very far apart in a thermodynamic sense.

However, just recently, a similar analysis15, 18 conducted for an ethylene plant in Texas19, 20 was made public; that investigation also included several catalyst test coupons exposed in SMR plants. Findings and conclusions from the two studies are similar.

Specifically, chromic oxide (Cr2O3) spontaneously forms a protective layer on the outside of the alloy metal tubes/coils exposed to high-temperature flue gas in the respective furnaces. This is accompanied by vaporization and disproportionation of the Cr2O3 into other chromium species. Energy-dispersive X-ray spectrometry (EDS)13, 21 shows chromium depletion from the SMR tube surface.

Chromium species are carried downstream and revert to their original form upon condensation on the relatively cooler SCR catalyst surface. Since they preferentially condense on fresh catalyst, contamination, with a darkening in color, progresses downstream from the catalyst inlet as time goes on. The contamination masks the active catalyst sites and gradually decreases its ability to promote the ammonia-NOx reactions. For uninterrupted operation, it is essential to predict how much catalyst must be installed to meet permitted NOx and ammonia levels at end-of-run, and to have catalyst replacement coincide with a scheduled turnaround.


Several essential points can be taken away from this discussion:

  • Steam cracking processes and SMR processes are different
  • Steam cracking and SMR burners/furnaces, where NOx is produced, are similar
  • NOx depends on fuel composition, combustion conditions and furnace temperature
  • AFT captures a wealth of fuel and air information
  • SMR burner-NOx correlation employing AFT can be extended to ethylene furnaces
  • SCR is utilized when burners alone cannot meet NOx limits
  • SCR has been successfully demonstrated for both types of plants
  • Chromium species in flue gas mask SCR catalyst and degrade performance over time. HP


This article is condensed from material contained in Paper 2a presented at the AIChE 2015 Spring Meeting in Austin, Texas, April 26–30, 2015, and Paper ENV-05-197 presented at the NPRA 2005 Environmental Conference in Dallas, Texas, September 18–20, 2005.


1 Kunz, R. G. and T. R. von Alten, “SCR treatment of ethylene furnace flue gas (a steam methane reformer in disguise),” Paper presented at the Institute of Clean Air Companies (ICAC) Forum ‘02, Houston, Texas, February 2002.
2 Suwa, A., “Operating experiences of SCR DeNOx unit in Idemitsu ethylene plant,” AIChE, Proceedings from the 13th Ethylene Producers’ Conference, New York, New York, 2001.
3 Kunz, R. G., D. D. Smith, N. M. Patel, G. P. Thompson and G. S. Patrick, “Control NOx from furnaces,” Hydrocarbon Processing, August 1992.
4 Kunz, R. G., D. D. Smith and E. M. Adamo, “Predict NOx from gas-fired furnaces,” Hydrocarbon Processing, November 1996.
5 Kunz, R. G., B. R. Keck and J. M. Repasky. “Mitigate NOx by steam injection,” Hydrocarbon Processing, February 1998.
6 Kunz, R. G., “Extension of NOx correlation to ethylene cracking furnaces,” Paper ENV-05-197, NPRA Environmental Conference, Dallas, Texas, September 2005.
7 Kunz, R. G., “Burner NOx from ethylene cracking furnaces,” AIChE, Proceedings from the 19th Ethylene Producers’ Conference, New York, New York, 2007.
8 Kunz, R. G., Environmental Calculations: A Multimedia Approach, John Wiley & Sons, Hoboken, New Jersey, 2009.
9 Gartside, R. J., P. R. Ponzi, F. D. McCarthy, S. G. Chellappan, P. J. Chapman and R. T. Waibel, “Commercialization of ultra-low-NOx burners for ethylene heaters,” AIChE, Proceedings from the 16th Ethylene Producers’ Conference, New York, New York, 2004.
10 Brandeis, B., “Air Products’ new environmental optimization and its significant cost savings for HGA industry” with slide presentation “Environmental solutions ... emissions optimizer and large-scale vortex—ultra-low-NOx burner,” Paper presented at Session B1 of Texas Technology Showcase, Houston, Texas, March 2003.
11 Kunz, R. G., “NOx, NOx, who’s there—steam cracker or SMR?” Paper 2a presented at the 2015 AIChE Spring National Meeting, Austin, Texas, April 2015.
12 Kunz, R. G., “SCR performance on a hydrogen reformer furnace,” Journal of Air & Waste Management Association, Vol. 48, 1998.
13 O’Leary, J. R., R. G. Kunz, and T. R. von Alten, “Selective catalytic reduction (SCR) performance in steam methane reformer service: The chromium problem,” Paper ENV-02-178, NPRA Environmental Conference, New Orleans, Louisiana, September 2002.
14 Ratan, S., N. Patel and B. Baade, “Driving hydrogen efficiency with an eye on the environment,” Hydrocarbon Engineering Reprint, February 2010.
15 Jensen, J. R., D. Salbilla and P. Lindenhoff, “A new standard of NOx control,” Hydrocarbon Engineering Reprint, October 2009.
16 Bay Area Air Quality Management District, “Final major facility review permit issued to Air Liquide Large Industries, US LP facility #B7419,” July 29, 2010, and “Air Liquide Rodeo Title V semi-annual monitoring report,” September 30, 2013.
17 Funahashi, K., T. Kobayakawa, K. Ishii and H. Hata, “SCR DeNOx in new Maruzen ethylene plant,” AIChE, Proceedings from the 13th Ethylene Producers’ Conference, New York, New York, 2001.
18 Jensen-Holm, H. and P. Lindenhoff, “Combating NOx from refinery sources using SCR,” Paper presented at 2nd Annual World Refining Technology Summit & Exhibition, Abu Dhabi, 2010.
19 De Haan, S., T. Garza and J. Hilbrich, “Performance, maintenance and environmental benefits of an integral SCR,” Paper 39d presented at the 2012 AIChE Spring National Meeting, Houston, Texas, April 2012.
20 Karrs, M. S. and P. S. Crepinsek, “Cracking heater retrofit with integral SCR,” Paper 86f presented at the 2008 AIChE Spring National Meeting, New Orleans, Louisiana, April 2008.
21 O’Leary, J. R., R. G. Kunz and T. R. von Alten, “Selective catalytic reduction (SCR) performance in steam methane reformer service: The chromium problem,” Environmental Progress, Vol. 23, Iss. 3, October 2004. 

The Author

Related Articles

From the Archive



{{ error }}
{{ comment.comment.Name }} • {{ comment.timeAgo }}
{{ comment.comment.Text }}