This is the old United Nations University website. Visit the new site at

Contents - Previous - Next

Table 1 Imputed efficiency of charcoal conversion (percentages)

Assumed weight of 1 cord Reported
conversion rates
c. 1840
35-40 bu/cord
(700-800 lb
c. 1875
45-50 bu/cord
(900-1,000 lb
c. 1910
60-65 bu/cord
(1,200-1,300 lb
2,500 lb (Ayres and Scarlott, 1952) 70-80 90-100 120-130b
3,000 lb (Riegel, 1937) 58.3-66.7 75-87.5 100-130b
4,000 lb (Schurr and
Netschert, 1960)
43.8-50.0 56.3-62.5 75.0-81.2

a. Assuming 12 per cent moisture.

b These figures are physically impossible.

The total amount of carbon moNoxide produced per unit of iron has dropped even more drastically, since charcoal manufacture a century or more ago was an inefficient process. Assuming cordwood to be 12 per cent moisture (H2O) and 88 per cent cellulose (C6H10O5)n, an ideal carbonization process would yield 0.39. (about 0.4) tons of charcoal (C) per ton of wood, with no emission of CO. However, in practice the yield of charcoal in the nineteenth century was less than this. A by-product wood distillation plant operated by the Ford Motor Co. (1936) yielded 649 lb of charcoal per (short) ton of dry scrap wood, or 0.3245 tons C per ton Fe (Riegel, 1937). In terms of charcoal output, this corresponds to an efficiency of 0.32/0.39, or 82 per cent, probably the highest level practically achievable. Yield efficiencies are calculated in table 1 for several different charcoal yields and wood density assumptions.

On the basis of these calculations it is fairly clear that the lower weight estimates are inconsistent with the higher estimates of charcoal output. This reduces the uncertainty. If conversion rates as high as 60 bu/cord were ever achieved the wood must have been both dry and unusually dense. We note that an average of 3,500 Ib/cord (12 per cent H2O) and a maximum conversion efficiency of 82 per cent would imply a charcoal output of 0.82 x 1,400 = 1,148 lb or 57.4 bushels of charcoal. This appears to be a reasonable figure. The 35-40 bu/cord conversion rate was reported as typical of current practice in the 1840s. Average yield in 1879 was 49 bu/cord, according to one source (Schurr and Netschert, 1960). Known technological improvements in the 1890s presumably increased the average yield to some degree. All things considered, we therefore adopt 3,500 (3,250-3,750) Ib/cord as the "best guess" of average wood density. This implies the historical trend in efficiency shown in table 2.

Table 2 Comparison of charcoal processes

Year Charcoal conver-
sion rate (bu/cord)
Charcoal conver-
sion efficiency (%)
Ratio of carbon input
(in wood) to carbon
output (in charcoal)
1840 35-40 50-57 1.75-2.0
1879 49 70 1.43
1937 57.4 82 1.22

Extensive use of anthracite as a substitute for charcoal in the iron industry came in the 1840s; it was made possible by the application of the hot blast to the iron-making process. Anthracite reduced fuel costs by almost half compared to charcoal, leading to its rapid adoption. Anthracite furnaces sprang up along the established coal trade routes close to urban areas, with the Lehigh Valley of Pennsylvania becoming the centre of the anthracite industry. By 1846, there were 43 anthracite blast furnaces in Pennsylvania and by 1856 there were 92. In the latter year Pennsylvania produced three times as much anthracite pig iron as charcoal iron, or approximately 300,000 tons (the national total was 394,509 tons) (Binder, 1974; Chandler, 1972).

According to historian Alfred D. Chandler, the combination of iron and coal "encouraged large-scale manufacturing of iron products for the first time. . ." (Chandler, 1972). Factories appeared for wire products, railway tracks, wheels, locomotives, steam engines, stoves, and agricultural equipment such as harvesters and ploughs.

Anthracite was also used in other industries requiring heat in the manufacturing process, such as glass and paper, baking, sugarrefining, and brewing, as well as in the processing of earthenware, plated ware, and chemicals (Chandler, 1972). While anthracite was to retain its importance in general industrial use and domestic heating for generations, its importance in the making of iron products was relatively short-lived, as shown in figure 4. Once the railroads had crossed the Appalachian Mountains in the 1850s, thereby making bituminous coal readily available, anthracite was rapidly displaced as a source of carbon in iron-making.

Combining all the above factors, we can now estimate the total production and emissions of carbon monoxide per ton of iron.

Fig. 4 Substitution of anthracite and coke for charcoal in iron-making in the US (Source: Temin, 1964)

In 1840, all iron was made with charcoal. The average consumption was 180 bushels (1.8 tons) of charcoal per ton of pig iron (94 per cent Fe), and each atom of carbon embodied in charcoal was accompanied by between 0.86 and 1.0 atoms of carbon in waste carbon moNoxide: we assume the median value of 0.93. A ton of pig iron, therefore, required the production of 1.8 x 0.93 x 28/16 = 2.93 (2.713.15) tons of carbon moNoxide. Essentially, all of this was presumably released into the atmosphere, either at the charcoal kiln or the blast furnace.

By 1880, only 13 per cent of US pig iron was made by the charcoal route; the remainder was based on coke or a mixture of coke and anthracite coal. An average "coke rate" of about 1.5 (1.4-1.6) can be assumed (fig. 3), while the charcoal furnaces of the time probably averaged around 140 bu/ton. Total CO production per ton of iron would thus be (0.13 x 1.4 x 0.43 + 0.87 x 1.5)(28/16) = 2.42 (2.352.50) tons, of which some fraction (probably less than half) was recovered for fuel use. Of course, total iron production in 1880 was much larger than in 1840. The coke rate dropped fairly steadily after 1880, mainly owing to the construction of newer and larger furnaces.

The fraction of CO recovered for use (as fuel) in the iron/steel industry is not easy to estimate. In 1840, essentially all the wood carbon not converted to charcoal (i.e. 43-50 per cent of the total) was emitted as CO. The first attempts to use blast-furnace gas as a fuel date from 1845, and the "bell and hopper" arrangement to permit furnace recharge without loss of furnace gases dates from 1850. But no practical success was met with until 1857 (McGannon, 1970). Unfortunately, blast-furnace gas is not a very good fuel. As it leaves the blast furnace, it consists of 24-28 per cent CO and 3-4 per cent H2O by volume. The rest is CO2 or N2. Blast-furnace gas has low heat value and a relatively low flame temperature (2,650 F). From the blast furnace it has 8-15 grains of dust per cubic foot, although a minimum cleanliness of 0.01 grains per cubic foot is now considered standard for most uses (McGannon, 1970). Thus, gas cleaning was necessary for practical recovery.

The first attempts to use electrostatic precipitators (ESPs) to clean blast-furnace gas occurred in 1919, but the process was not perfected for this application until 1930 (White, 1957). Blast-furnace gas recovery for in-plant fuel use can be assumed to have been negligible up to 1910, and no more than modest until after gas-cleaning technology had been perfected. However, with the availability of costeffective technology, carbon moNoxide recovery within large integrated iron and steel works was well advanced by 1937 and essentially complete by 1955. However, recovery is far less efficient in smaller specialty plants or other industries.

Hence, in the absence of better data it seems safe to suggest that most of the carbon moNoxide produced in the iron/steel industry was not recovered until well into the twentieth century. We assume an increasing recovery fraction of around 20 per cent (15-25) in 1920, rising to 45 per cent (35-50) by 1937, reaching 75 per cent (60-80) by 1950 and 85 per cent (80-90) in 1960 (fig. 5). Even today, a considerable amount of CO is vented without flaring, principally by iron foundries, although the overall percentage of emissions is now rather low.

Petroleum refineries were also major sources of carbon moNoxide at one time, mainly from catalyst regeneration (after 1900) and decoking of thermal distillation, cracking, and reforming units. In the 1950s, uncontrolled Los Angeles refineries generated about 20 kg of CO per ton of crude oil processed (Elkin, 1968) or a coefficient of 0.02. However, controls introduced by 1966 cut this down to about 1.2 kg/ton (0.0012). Most US refineries have since introduced even more rigorous controls, not so much to eliminate CO emissions as to eliminate the associated hydrocarbons, especially toxic PAHs. We assume that CO emissions per ton of crude oil processed in 1920 were significantly higher (0.05 to 0.10 tons/ton), mainly as a result of the decoking of batch-type thermal crackers.

The EPA attributes around 6.3 Tg of CO annually (1980) to industrial offgases, down from 9.0 Tg in 1970 (USEPA, 1986). However, a comprehensive survey of industrial sources of by-product CO carried out in 1977 indicated a total production of 142 million tons (128 Tg), of which 117 million tons were attributable to iron and steel manufacturing, and 14.5 million tons were generated by the petroleum industry (Rohrmann et al., 1977). Only 63 million tons were actually recovered as fuel (see table 3). Of the remainder, 57 million tons were classed as "not recoverable," on the grounds that it was largely oxidized to CO2 at furnace walls prior to emission, and 22 million tons were identified as "flared or exhausted." As of the mid-1970s, some 17.8 million tons were still flared by the iron and steel, petroleum and carbon black industries (Rohrmann et al., 1977).

These figures imply that 4.2 million tons were vented by these three industries. It is virtually certain that some of the 57 million tons "not recoverable" was actually vented as CO, though in dilute form. The question is: how much? If 10 per cent of the "not recoverable" fraction were emitted, it would mean total industrial offgas emissions of 9.9 million tons. This is close to the EPA estimate, and seems plausible. However, if 20 or 30 per cent of the "unrecoverable" CO were actually emitted as such, a much larger total would result. Thus, the EPA emissions estimates are more likely to be too low than too high.

Fig. 5 CO production and emission per ton of pig iron

Table 3 Estimated annual production of by-product CO,1975


1,000 tons/yr/no. of sites

vol (%)
from process
Used entirely as
fuel at present
Flared or
Iron and steel          
Sinter 1,090 (41) 1,090 (41) 0 1,090 (41) 0.35
Blast 58,280 (53) 58,280 (53) 43,898 (53) 12,382 (53) 22.5
BOFa 28,126 (36) 527 (36) 0 527 (36) 30
BOFb 3.781 (2) 3,781 (2) 3,781 (2) 0 50
Electric 25,751 (125) 294 (125) 0 294 (125) 0-6
Ferroal loys          
FeMn 976 (9) 617 (1) 543 (1) 617 (1) 40
FeCr 413 (10) NA 0 NA  
FeSi 1,484 (15) 300 (3) 78 (1) 300 (3) 90
SiMn 148 (8) 44 (g) 0 44 (1)  
Other alloys 250 (11) NA 0 NA 65
Coke ovens 2,439 (63) 2,439 (63) 2,439 (63) 0 6
Silicon products 370 (10) NA 0 NA  
Aluminium 1,695 (32) 244 (5) 0 224 (5) 30
Phosphorus 1,316 (9) 1,316 (9) 1.184 (9) 132 (9) 98
Titanium dioxide 275 (8) 275 (8) 0 275 (8) 15
Carbon black 2,275 (29) 2,275 (29) 33 (1) 2,242 (29) 6
Calcium carbide 183 (5) 165 (3) 65 (2) 100 (3) 75
Lead and tin 363 (7) 363 (7) 0 363 (7) <1
Zinc (thermal) 162 (2) 162 (2) 142 (2) 20 (1) 80
Petroleum refiring 14,561 (149) 14,561 (149) 10,921 (112) 3,640 (37) 8
Kraft paper pulping 160 (116) 160 (116) 0 160 (116) <1
Total 142,098 84,893 63,084 21,709  

NA = negligible CO available for recovery under present process conditions. a. Without suppressed combustion. b. With suppressed combustion.

It is evident that emissions from two major industrial CO sources (iron and steel, petroleum) have dropped sharply in recent decades. This is partly due to increased process efficiency and partly to concern for emissions control per se. Emissions from the paper and pulp, carbon black, petrochemical, ferroalloy, and non-ferrous metals industries have also dropped, though probably not as dramatically, since these industries are less concentrated. We can say, however, that in comparison with the petroleum and iron and steel sectors, they have not, until recently, contributed more than a small fraction of the total industrial emissions.

One other historical, but not contemporary, source of industrial CO emissions is worth mentioning: manufactured or "town" gas. Although the production of manufactured gas began early in the nineteenth century, it reached its peak from about 1880 to 1950, when it was largely displaced by natural gas. Manufactured gas was most widely used for lighting, as a cooking fuel, and for heating.

An authoritative study of the manufactured gas industry estimates that approximately 15 trillion cubic feet of gas was manufactured in the United States from 1880 to 1950. The number of sites in the nation during this period ranged from a high of about 1,000 in 1890 to a low of 200 in 1950. Size of plant increased considerably during this period, with the highest gas production of 365,000 million cubic feet occurring in 1930, from 737 plants, and the second highest in 1950, with 331,000 million cubic feet of gas from 194 plants. These plants were widely distributed throughout the nation, although New York, New Jersey, Massachusetts, and Pennsylvania had the highest production.

Manufactured gas was produced from coal by a variety of methods. The most common form of manufactured gas was so called water gas or "blue gas," produced by reacting coal or coke with steam at 1,000 to 1,400 C, yielding a gas rich in hydrogen and carbon moNoxide with a heating value of about 300 BTU per cubic foot. On the West Coast manufactured gas was generated from petroleum in a process that thermally cracked the oil into a gaseous product known as "oil gas" (ERT, 1984).

This process was a major user of coal and coke in the United States, especially during the period 1900-1950. Losses of CO in the manufacturing process and in distribution were never accurately assessed, but could scarcely be much less than 2-3 per cent, and probably reached 5 per cent or more. Thus, in comparison with other industrial sources of CO at the time, this contribution was rather low.

Carbon monoxide emissions coefficients and estimates

Table 4 gives the USEPA (1986) estimates of CO emissions for the United States by decade from 1940 to 1980, by source category. As noted in the table itself, some of the earlier industrial estimates seem likely to be significantly too low, for reasons discussed in the text. Our estimates are based on known characteristics of combustion processes and metallurgical reduction processes and (admittedly crude) estimates of CO recovery/disposal efficiency in the past. However, we conclude that the EPA estimates for uncontrolled industrial emissions are more likely to be too low than too high and that their historical estimates are certainly too low. Table 5 gives our assumed emissions coefficients.

Table 4 EPA estimates of CO emissions 1940-1980 (teragrams/yr)

Source 1940 1950 1960 1970 1980
Highway vehicles 22.0 33.1 46.5 62.7 45.3
Aircraft 0.0 0.8 1.6 0.9 1.0
Railroads 3.7 2.8 0.3 0.3 0.3
Vessels 0.2 0.2 0.6 1.1 1.4
Off-highway 3.4 6.7 8.0 6.8 4.7
Mobile source total 29.3 43.6 51.0 71.8 52.7
Electric power 0.0 0.1 0.1 0.2 0.3
Industrial boilers 0.4 0.5 0.6 0.7 0.6
Commercial/institutional heat 0.1 0.1 0.0 0.1 0.1
Residential heat 15.4 10.7 6.3 3.4 6.4
Stationary combustion total 15.9 11.4 7.0 4.4 7.4
Chemicals 3.8 5.3 3.6 3.1 2.0
Petroleum 0.2a 2.4a 2.8 2.0 1.6
Iron and steel mills 1.5 1.1a 1.3 1.6 0.9
Primary aluminium 0.0 0.1 0.3 0.6 0.8
Iron foundries 1.0 1.4 1.0 1.1 0.3
Pulp mills 0.1 0.2 0.3 0.6 0.7
Industrial total 8.6a 10.5a 9.3 9.0 6.3
Incineration 2.0 2.5 2.5 2.7 1.2
Open waste burning 1.3 1.8 2.6 3.7 1.0
Forest fires 22.8 12.8 6.7 5.1 6.9
Other burning 3.7 3.7 3.3 2.1 0.7
Waste and miscellaneous total 29.8 20.8 15.1 13.6 9.8
Grand total 81.6 86.3a 88.4a 98.8 76.2

Source: USEPA, 1986. a. Almost certainly too low (see text).

Table CO emission coefficients (tons CO/ton fuel)

Process Fuel 1800 1860 1890 1920 1950 1980
Stationary fuel combustion
for residential heating
and cooking

Anthracite coal


Natural gas

0.05 0.09 0.072 0.054 0.036 0.063
0.08 0.08 0.08 0.08 0.08 0.08
Boiler fuel (all kinds)

Petroleum refining runs to


Coal and oil 0.005 0.005 0.005 0.004 0.003 0.002
  - - - 0.02 0.02 0.023
Auto transportation Gasoline - - - 0.34 0.34 0.16
Iron and steel manufacturing

(inc. charcoal)

  8.11 2.15 1.10 0.25 0.10 0.02


Contents - Previous - Next