Energy used to manufacture agricultural chemicals

In the original version of the model, the estimates of the energy requirements for fertilizer were based on a review of the estimates of others. I have replaced these with estimates derived from primary survey data, from the Census and the EIA, on energy input and product output at the mining, transport, and manufacturing stages.

According to Bhat et al. (1994), the fertilizer industry researched, developed, and adopted significant energy saving techniques beginning in the late 1970s and continuing through the mid 1980s (no doubt in reaction to the energy crises of the 1970s, and high oil prices of the early 1980s). Bhat et al. (1994) themselves make detailed estimates of the energy inputs to the lifecycle of fertilizers, based on 1987 data on energy use in fertilizer manufacture, and earlier data regarding energy use at other stages of the lifecycle. However, because the recent primary data from the EIA and the Census are about a decade more current than the data of Bhat et al. (1994), and reflect all of the energy-efficiency measures adopted through the mid 1980s, I have done an original analysis of energy inputs into the fertilizer lifecycle, rather than rely on the more detailed but less current estimates of Bhat et al. (1994).

Although the energy-input/product-output estimates are not particularly complex, neither are they completely straightforward. There are in general four kinds of problems that must be faced. First, primary data are not available for every stage of all of the chemical lifecycles. Second, the survey data that are available have deficiencies. Third, the scope of the energy input data and the scope of the product output data rarely match precisely. Fourth, because fertilizer use is reported in nutrient tons, as opposed to material tons, one must estimate the relationship between material tonnage and nutrient tonnage. These issues are discussed in more detail as they arise in particular contexts.

The lifecycle of nitrogen, phosphate, and potash fertilizers

The major fertilizers, or "macro nutrients," are nitrogen, expressed in terms of elemental N; phosphate, expressed in terms of P2O5; and potash, expressed in terms of K2O. The lifecycle of potash is the simplest. Virtually all of the potash used in agriculture is potassium chloride (Taylor, 1994), KCl, which is a refined product of potassium mining establishments (Bureau of the Census, 1992 Census of Manufacturers, Agricultural Chemicals , 1995). Hence, the energy lifecycle of potash comprises energy inputs to:

Potash:

• mine production of KCl

• transportation of KCl

• mixing of fertilizer for sale

The lifecycles of phosphate fertilizers are more complex. There are several different kinds of phosphate fertilizer (Taylor, 1994; Bureau of the Census, Current Industrial Reports, 1997), and several major inputs into the manufacturing process (Bureau of the Census, 1992 Census of Manufacturers, Agricultural Chemicals , 1995). I will treat the two major inputs, phosphate rock and sulfur, in detail, and then scale these calculated results to account for other inputs. Thus, the energy lifecycle of phosphate fertilizer comprises energy inputs to:

Phosphate fertilizer:

• mine production of phosphate rock

• production of sulfur

• transportation of phosphate rock and sulfur

• manufacture of fertilizer

• production of materials other than phosphate rock and sulfur

• mixing of fertilizer

• transport of finished fertilizer

The lifecycles of nitrogenous fertilizers are the most complex of all -- too complex, in fact, to be characterized in detail. Consequently, I will estimate energy use at the major manufacturing stage, and then apply a scaling factor to account for energy inputs to the materials input to the manufacturing stage:

Nitrogen fertilizer:

• manufacture of fertilizer in SIC 2873

• production of materials input to SIC 2873

• mixing of fertilizer

• transport of finished fertilizer

Potash

Mineral mining. The 1992 Census of Mineral Industries, Fuels and Electric Energy Consumed (Bureau of the Census, 1997) reports energy consumed by industries that mine fertilizer and chemical minerals: SIC 1474, potash, soda, and borate mining; SIC 1475, phosphate rock mining; and SIC 1479, fertilizer and chemical mineral mining not elsewhere classified. The Bureau of the Census, 1992 Census of Mineral Industries, Chemical and Fertilizer Mineral Mining, (Bureau of the Census, 1995) reports the tons of minerals shipped from the same industries, in 1992. With these two data sources, one can estimate the amount and kind of energy use per lb of fertilizer mineral shipped.

There are three considerations in the estimation of this BTU/lb measure. First, the Census data on energy use are not complete: some data are withheld to avoid disclosing information for individual companies, and some energy use (called "undistributed fuels") is reported as dollar expenditure rather than physical energy quantity. I have filled in the withheld data so that all of the individual cells add to the higher level totals shown, and have assumed that the undistributed energy comprises the same mix of fuels as the distributed energy.

Second, and most seriously, much of the data on tons of mineral shipped in 1992 were withheld in order to avoid disclosing data for individual companies. I have estimated the withheld data on the basis of value of shipments, shipments in prior Census years, other data on mineral production, and my judgment.

Third, the energy required to mine a ton of potash specifically might be more or less than the energy required to mine a ton of potash, soda, and borate mineral on average. However, I have no basis for assuming that it is different.

It appears that mining energy has been declining dramatically. I will assume that this trend continues.

The estimates of BTU/lb-fertilizer-mineral must be converted to the BTU/lb-fertilizer-nutrient provided (as N, P2O5, and K2O). This is done by multiplying BTU/lb-fertilizer-mineral by the ratio of the weight of fertilizer minerals to the weight of N, P2O5, and K2O nutrient provided. This ratio is discussed below.

A portion of the energy used in industry 1481, which provides services in support of all industries in major SIC group 14 (mining and quarrying of all non-metallic, non-fuel minerals) should be assigned to the mining of fertilizer and chemical minerals. In 1992, SIC 1481 consumed 1% of the energy consumed by all non-metallic, non-fuel mineral-mining industries in SIC 14. On the assumption that the energy used in SIC 1481 should be allocated to the mining industries of SIC 14 in proportion to energy consumption, I scale the energy use in SICs 1474, 1475, and 1479 by 1.01.

These data, along with my assumptions regarding energy intensity per mode, can be used to estimate the average BTUs required to transport a lb of any fertilizer mineral. Note that the transport energy thus calculated is expressed per lb of mineral, not per pound of finished fertilizer nutrient. Because fertilizer nutrient is a fraction of the total material, BTU/lb-fertilizer-nutrient will exceed BTU/lb-material. This is addressed below.

Fertilizer mixing. In the Standard Industrial Classification, which serves as the basis of the EIA’s and the Census’ estimates of manufacturing energy consumption by industry, there is a separate industry engaged in "fertilizer mixing only" (SIC 2875). This industry includes establishments "primarily engaged in mixing fertilizers from purchased fertilizer materials" (Office of Management and Budget, 1987, p. 147). I treat this industry as if it were part of the manufacturing or mining stage: I estimate the total energy in the industry, and allocate the total to each of the three major fertilizers (nitrogen, phosphate, and potash).

Unfortunately, the EIA’s Manufacturing Energy Consumption survey does not cover SIC 2875. However, the 1992 Census of Manufactures, Agricultural Chemicals (Bureau of the Census, 1995) does report total electricity consumption (289 million kWh), and total expenditures on fuels in SIC 2875. Comparing these expenditures with the reported expenditures on fuels in SICs 2873 and 2874, and then refering to the EIA’s estimates of fuel consumption in SICs 2873 and 2874, I estimate that SIC 2875 consumed about 10 trillion BTUs of fuel in 1992, mainly as natural gas. This is a relatively small amount.

Ratio of material weight to nutrient weight. Fertilizer application is reported not in tons of actual material applied, but in tons of nitrogen (N), phosphate (as P2O5), and potash (as K2O) nutrient. However, the energy required to mine fertilizer minerals, and transport fertilizer minerals and products, is related to and expressed in terms of the total material weight. To express the transport and mining energy per unit of N, P2O5, and K2O provided, we must multiply the energy/total-material-ton by the ratio of material tons to nutrient tons.

The case of potash is simple, because virtually all potash fertilizer is potassium chloride, KCl, a basic mineral produced from potassium mining with essentially no chemical refining. Two moles of KCl (molecular weight 74.55 g/mole) are needed per mole of K2O (molecular weight of 94.2); hence, 1.58 tons of KCl (74.55 . 2 ÷ 94.2) provides 1.0 ton of K2O.

Phosphate

Mining of phosphate rock. With the Census’ data on energy use (1992 Census of Mineral Industries, Fuels and Electric Energy Consumed, Bureau of the Census, 1997), and net shipments (Bureau of the Census, 1992 Census of Mineral Industries, Chemical and Fertilizer Mineral Mining, 1995) in SIC 1474, I estimate the amount and kind of energy use per lb of phosphate rock shipped. See the discussion above regarding mining of potash, soda, and borate minerals, in the "Potash" section, for further details. Note that for two reasons, the BTU/lb estimate for phosphate rock is more accurate than the BTU/lb estimate for potash. First, SIC 1475 comprises only phosphate rock mining, whereas SIC 1474 includes borate and soda mining as well as potash mining. Second, the Census estimates of shipments of phosphate rock are complete. Thus, the second and third caveats above regarding the data on potash mining do not apply here.

I assume that the trend of declining energy use continues, albeit less dramatically.

Transport of fertilizer mineral. The Bureau of the Census’ 1993 Commodity Flow Survey (Bureau of the Census, 1996) reports tons and ton-miles of chemical and fertilizer minerals (STCC 147: potash, soda, borate, phosphate rock, sulfur, and other chemical and fertilizer minerals) shipped by rail, ship, truck, and other modes, in 1993. See the discussion in the "Potash" section for further details.

Fertilizer manufacture. In 1994, SIC 2874 consumed about 46 trillion BTU of primary energy for all purposes, almost entirely in the form of natural gas (EIA, Manufacturing Consumption of Energy 1994, 1997; electric energy counted at 3412 BTU/kWh). In 1992, the value of the primary products (code 2873 ---) of this industry was 91% of the total value of all products (primary + secondary) (Bureau of the Census, 1992 Census of Manufacturers, Agricultural Chemicals , 1995). Hence, I assign to the production of phosphate fertilizer 91% of the reported total energy consumption in SIC 2873 in 1994.

In 1994, 14.2 million tons of P2O5 in phosphate fertilizer (product codes 2874 ---) were shipped in all industries (Bureau of the Census, Current Industrial Reports , 1997). In 1992, 92% of all phosphate fertilizers were produced in SIC 2874 (Bureau of the Census, 1992 Census of Manufacturers, Agricultural Chemicals 1995). Hence, I assume that SIC 2874 produced 14.2 . 0.92 . 2000 = 26.1 billion lbs of P2O5 in 1994.

The resulting energy-in/product-out ratio is about 1,600 BTU/lb. For two reasons, this is a lower bound on total energy requirements for the manufacture of nitrogen fertilizer. First, it does not include the energy required to produce the materials other than sulfur and phosphate rock that are input to SIC 2873. Second, the Census’ estimate of the weight of shipments, which is the denominator of the BTU/lb estimate, probably overstates the net output of the industry. The Census estimate of total shipments excludes material produced and consumed in the same plant, but includes inter-plant transfers. To the extent that material transferred from plant A to plant B is counted once as an inter-plant transfer, and again as part of the finished output of plant B, the true net output of finished products from SIC 2874 will be overestimated. With these considerations, it does not seem unreasonable to that the true BTU/lb manufacturing energy requirement is 10% higher than the figure just calculated. I assume that the average energy requirement, in BTU/lb, declines slightly.

Sulfur. The phosphate lifecycle includes lifecycle emissions from the sulfur input to the production of phosphate fertilizer. Emissions from the production of sulfur depend on the process; presumably, it takes less energy to recover sulfur from waste streams than to produce sulfur from virgin ore. I assume that 70% of sulfur is recovered from waste streams, at a cost of very little GHG emissions. Emissions from the recovery of the 30% from virgin ore are estimated on the basis of the Census energy use data (1992 Census of Mineral Industries, Fuels and Electric Energy Consumed, Bureau of the Census, 1997).

Emissions from sulfur transport are estimated on the basis of data reported in the Census’ 1993 Commodity Flow Survey (Bureau of the Census, 1996). See the discussion in the "Potash" section for further details.

Fertilizer mixing. See the discussion in the "Potash" section.

These data, along with my assumptions regarding energy intensity per mode, can be used to estimate the average BTUs required to transport a lb of any fertilizer product. Note again that the transport energy is expressed per lb of total product material (e.g., NH3), not per pound of finished fertilizer nutrient (e.g., N). Because fertilizer nutrient is a fraction of the total material, BTU/lb-fertilizer-nutrient will exceed BTU/lb-material. This is addressed next.

Ratio of material weight to nutrient weight. In the case of phosphates, three ratios are of interest:

1) the ratio of the weight of phosphate rock mined and shipped to the weight of P2O5 provided;

2) the ratio of the weight of sulfur input to manufacturing to the weight of P2O5 provided; and

3) the ratio of the weight of finished fertilizer product shipped to the weight of P2O5 provided.

1). Phosphate rock. Three different estimates indicate the ratio of phosphate-rock weight to P2O5 weight is on the order of 3:1. First, the TRW study cited in Appendix K of Volume 2 (DeLuchi, 1993) indicates a ratio of 3:1. Second, the Encyclopedia Britannica states that typical phosphate rock beds contain about 30% P2O5 . Third, the rock-input/P2O5-output ratio for industry 2874, phosphate fertilizers, is almost 3:1. In 1992, this industry consumed 34 million tons of phosphate rock (Bureau of the Census, 1992 Census of Manufactures, Agricultural Chemicals, 1995). In 1992, all industries shipped about 14 million tons of phosphoric acid and superphosphate fertilizer, as P2O5 weight (Bureau of the Census, Current Industrial Reports, 1997). According to the 1992 Census of Manufactures, Agricultural Chemicals (Bureau of the Census, 1995), SIC 2874 accounted for 92% of the total value of shipments of phosphoric acid and phosphate fertilizer. Hence, the rock-input/P2O5-output ratio for industry 2874 was 34/(14 . 0.92) = 2.6.

2). Sulfur. In 1992, the weight of sulfur input to SIC 2874 was 25% of the weight of phosphate rock. If the phosphate-rock/P2O5 weight ratio is 3.0, then the sulfur/P2O5 weight ratio is 0 .75:1.

3). Finished fertilizer. The ratio of the gross weight of superphosphates and other phosphate fertilizer shipped to the weight of the P2O5 content shipped, in 1994, was 1.88 (Bureau of the Census, Current Industrial Reports, 1997). (The ratio of the weight of pure phosphoric acid to P2O5 content is 1.4.) I will assume a value of 1.90 for this industry.

Nitrogen

Fertilizer manufacture. A lower bound on the energy required to manufacture nitrogen in fertilizer can be estimated on the basis of aggregate energy-in/product-shipped data for SIC 2873, nitrogenous fertilizers. The primary products of this industry are anhydrous ammonia, nitric acid, ammonium nitrate, ammonium sulfate, nitrogen solutions, urea, and nitrogen organic fertilizers [Office of Management and Budget, 1987].) These primary fertilizer products have a seven-number code, the first four digits of which are the SIC code of the nitrogen fertilizer industry (2873).

The EIA reports total energy consumed in SIC 2873, and the Census reports total production of fertilizer products (code 2873 --). However, these two figures by themselves do not give the energy-in/product-shipped ratio of interest, because SIC 2873 produces a minor amount of non-fertilizer ("secondary") products, which have a product code that does not begin with 2873, and some industries other than 2873 produce nitrogen fertilizer products, with code 2873 ---.

In 1994, SIC 2873 consumed over 600 trillion BTU of primary energy for all purposes, almost entirely in the form of natural gas (EIA, Manufacturing Consumption of Energy 1994, 1997; electric energy counted at 3412 BTU/kWh). In 1992, the value of the primary products (code 2873 ---) of this industry was 94% of the total value of all products (primary + secondary) (Bureau of the Census, 1992 Census of Manufacturers, Agricultural Chemicals , 1995). Hence, I assign to the production of nitrogen fertilizer 94% of the reported total energy consumption in SIC 2873 in 1994.

The Bureau of the Census Current Industrial Reports (1997) series shows the total production and total shipment of nitrogen fertilizers (product codes 2873 ---) in short tons in 1994. Multiplying total shipments by the nitrogen weight fraction, for each product class (see below) results in a total of 16.6 million short tons of N produced in 1994. In 1992, 79% of nitrogenous fertilizers (products with a code of 2873 --- ) were produced in SIC 2873 (Bureau of the Census, 1992 Census of Manufacturers, Agricultural Chemicals 1995). Hence, I assume that SIC 2873 produced 16.6 . 0.79 . 2000 = 26.2 billion lbs of N in 1994.

The resulting energy-in/product-out ratio is about = 22,000 BTU/lb. However, for three reasons, this is a lower bound on total energy requirements for the manufacture of nitrogen fertilizer. First, it does not include the energy required to manufacture the materials that are input to SIC 2873. Second, it does not include any energy required for further processing of the output of SIC 2873. Third, the Census’ estimate of the weight of shipments, which is the denominator of the BTU/lb estimate, probably overstates the net output of the industry. (See the discussion above in regards to phosphate fertilizers.) With these considerations, it does not seem unreasonable to that the true BTU/lb manufacturing energy requirement is 20% higher than the figure just calculated. I assume that the average energy requirement, in BTU/lb, declines slightly over the projection period.

Energy mix. The EIA manufacturing energy consumption survey shows that 97.4% of the energy used in SIC 2873 in 1994 was natural gas, 2.1% was electricity, and 0.5% other fuels (EIA, Manufacturing Consumption of Energy 1994, 1997). It is likely that energy uses at some of the other stages of the nitrogen-fertilizer fuelcycle (such as mining and transport) involve relatively more diesel fuel. Hence, I assume an overall breakdown of 96% natural gas, 2% electricity and 2% diesel fuel.

Fertilizer mixing, and transport of fertilizer product. See the discussion in the "Potash" section.

Ratio of material weight to nutrient weight. The ratio of material weight to nitrogen weight can be estimated as:

With this formula, and assuming that 100%-N nitrogen solutions have a 2:1 material/nitrogen ration (the Census reports the weight of nitrogen solutions in terms of 100% N), I estimate that MNR = 1.82 for 1994. One gets a similar result, 1.89, by dividing the total amount nitrogen materials consumed in the U. S. in 1993 (21.5 million tons) by primary nitrogen consumption in 1993 (11.4 million tons), as reported by Taylor (1994). I will assume a value of 1.90.

Energy used to manufacture pesticides

Turhollow’s (1997) update of Bhat et al. (1994) is the most comprehensive and recent estimate of energy embodied in agricultural pesticides. Bhat et al. (1994) proceed in four steps. First, they identify the pesticides most commonly used on major field crops in the U. S., and then group them according to their chemical structure. Next, they estimate the manufacturing energy required for each kind of pesticide. (These estimates, in GJ/Mg, are taken from Green’s (1987) oft-cited contribution to the series Energy in World Agriculture). Third, they calculate the application-weighted average energy content (GJ/Mg) for herbicides, insecticides, and fungicides. Fourth, they add energy for formulation, packaging, and transport. Turhollow’s updated result is about 105,000 BTU/lb for all pesticides.

In the chemical industry as a whole (SIC28), and in the miscellaneous organic chemicals industry, the breakdown of primary energy use is about 45% natural gas, 30% LPG, 8% electricity, 7% coal, and 10% oil (EIA, Manufacturing Energy Consumption Survey, 1994). However, Turhollow (1997) indicates a much larger share for oil. My assumptions are shown in Table XV.

Summary of estimates and assumptions regarding energy in the lifecycle of agricultural chemicals

Miscellaneous