Conserving energy in the production of aluminum

Resources

and Energy

1 (1978) 2142.

0 North-Holland

CONSERVING PRODUCTION

Company

ENERGY IN THE OF ALUMINUM

Richard

A. CHARPIE

Sloan School of Management, M.I.T,

Paul

Pubhshing

Cambridge, MA 02139, U.S.A

W. MacAVOY

School of Organuatlon and Management,

Yale Unirersity, New Haven, CT 0651 I. U.S.A.

The questton IS whether signiticant reductions in energy utrlizatton can be expected from the aluminum industry. It is of interest also to question whether such reducttons if any would come from the application of government ‘conservation’ policy, and/or from market responses of aluminum producers to higher energy prices. The extsting research hterature IS summartzed and extended by use of a simple engineering-economtc model of energy utilization. The answers tf preliminary appear to be in the negative, except for policies designed to sttmulate recyclmg of alummum scrap.

1. Introduction There has been considerable policy-related discussion of ‘conservation’ of energy in industry since the oil embargo of 1973-1974. Much of this discussion has been confusing because of failure to distinguish between ‘conservation’ brought about by policy changes and that resulting from market-determined price increases working on consumers to reduce their demands. Conservation of energy in aluminum (Al) could be the result of: (1) new methods of making aluminum, (2) government intervention to mandate decreases in energy consumption, and (3) changing production practices to reduce usage as a result of increased fuel prices and of concern or awareness of possible future restrictions on energy utilization. Some part of (1) and (3) could occur without the application of public policy, as the usual market demand reaction in face of higher energy prices. But not all energy reduction strategies are cost-effective and specification of (1) and (2) may give some clues regarding where government conservation policies would be most likely implemented.’ ‘As Fellner (1951) points out, the correct definitton of cost-effectiveness is not Immediately obvtous. Though a new technology may produce output at a lower average total cost than the extstmg technology, tt does not follow that old technology plants should be scrapped and replaced by new technology plants Rather, the average total cost of the new technology must be less than the average variable cost of the old technology to Justify Immediate replacement of the old technology plants. Nonetheless, all new capacity will be bum using the new technology because of tts lower average total cost. This simple analysts is complicated when the market structure of the Industry under consideranon is not competttive. For more detatls, see Fellner (1951)

22

R.A. Charpie

and P.W. MacAooy,

Conservrng

energy

rn the production

of alummum

This industry has become a target of conservation strategies, in part because large amounts of energy are used in primary aluminum production. For example in 1970 aluminum producers consumed approximately four percent of the electrical energy generated in the U.S.’ Recent energy accounting studies aimed at identifying opportunities for conservation indicate that the energy consumed in producing one pound of primary aluminum product from raw materials is very large as far as industry-toindustry comparisons go. The activities alleged to offer savings are customarily divided into three steps. The first step in aluminum production is the formation of alumina (Al,O,) from mixed bauxite ores. In the Bayer process, bauxite ore is digested in a hot caustic soda solution, which permits the precipitation of an alumina solution which is then calcined to remove water. The mining process consumes approximately 2,000 British Thermal Units/pound (BTU/lb.) Al, and the Bayer process consumes 12,000 BTU/lb. Al.3 In the second step, alumina from the Bayer process is dissolved in a molten cryolite and the mixture is placed in an electrolytic cell where the aluminum ions are separated from the oxygen ions. This process consumes 1,OOOlbs. carbon and 100lbs. cryolite per ton of aluminum produced; the carbon anode preparation represents an energy input of 2.800BTU/lb. A1.4 The direct electricity consumption amounts to’7.8 kilowatt hours/pound (kWh/lb.) (81,90OBTU/lb.). This is an industry average figure; most of the smeltmg pots operate in the range of 6.5-9.0 kWhjlb. (68.200there is ancillary electricity usage of 94,500 BTU/lb.).’ In addition, 0.6 kWh/lb.(j These figures are direct current powers, the industry convention. and must be distinguished from alternating current which is three to five percent higher. The energy in the third or fabrication step is more difficult to determine in the accounting scheme. This is because the energy used for fabrication varies with the type of final product -it takes more energy to make a unit of foil than it takes to make the same unit of plate. Typical data from industry sources indicate that from 15,000 to 30,000 BTU/lb. of fabricated product can be expected. This paper will use a figure of 20,OOOBTU/lb. for succeeding calculations. although this is difficult to justify on any sort of theoretical basis. Thus the energy accounting calculations are: Mining-2,000 BTU/lb.; Bayer 12,000 BTU/lb.; Anodes- 2.800 BTU/lb.; Smelting-91.700 BTU/lb.; Fabrication - 20,000 BTU/lb.; Total -128,500 BTU/lb. using a conversion 2Flemmgs et al t 1974. p. 1) ‘Conference Board (1974, p. 531) ‘ConTerence Board (1974, p. 533)-the welghted Soderberg capacity = 33 “,,, prebake capacity = 67 y,. %yftopolous et al. (1974. p. 68). “FEA Task Force Report (1974, pp. 684).

average

of these

data

IS computed

usmg

R.A.

Chorpie and P.W: Muc.410~. Conserrmg

energ)

m the producrron

of aluminum

3

factor of 10,500 BTU/kWh (the 1971 national average utility heat conversion ratio). These energy accountings are inexact and biased downwards inasmuch as the entire network of energy use has not been traced through. For example, the calculations exclude the energy content of the cryolite and many other indirect energy inputs into the aluminum manufacturing process. In addition the allocation of energy demand to fabrication is notably imprecise. Recent experience in trying to identify energy inputs in the milling of superalloys to sheet, bar, and wire indicates that any numbers which are quoted for the fabrication energy demand of final aluminum mill products Finally the amounts of secondary will have large margins of error.’ aluminum used in process significantly affect the total energy used, as is discussed later. Nonetheless, the estimate provides a measure of magnitude for comparisons between various products and to point out where the major energy applications occur. The major source of energy in aluminum processing is electricity. Of this source, 67 percent is purchased and the remainder generated by the industry itself.* Generation of electricity is by hydro, gas, and coal sources. In the period through 1985 the aluminum companies will probably not build additional generating capacity due to large capital costs and long payback periods, and the Conference Board Report argues that additional electricity will be purchased.’ However, representatives of the Bonneville Power Administration have indicated that no large blocks of firm power will be made available to the aluminum industry in the Pacific Northwest from the Federal system. As a result, many industry representatives argue that participation of aluminum producers in the construction of joint electrical generation facilities may well occur. Besides electricity, natural gas provides 30 percent of the energy inputs into the aluminum manufacturing process.” Though some of this gas is used indirectly to generate electricity at the plant sites, the major portion is used in fabrication. One industry source estimates that 20 percent of the gas presently used is ‘essential gas’; i.e., it cannot be replaced by any other fuel. Direct uses of coal and oil are minor sources in the present supply system although a shift towards oil and away from gas is now taking place to achieve more balance in energy supply and to avoid the supply interruptions associated with natural gas. In the short term, coal consumption is expected to remain stable due to environmental constraints and in spite of large coal holdings by aluminum companies. In the long run, coal gasification may influence the energy supply mix.

‘See Charpie (1975). ‘The carbon Input IS energy-Intensive, usage are preferred. ‘Conference Board (19!4, p. 544). “Ibid, p. 545

so that the processes

low in both

energy

and carbon

24

R.A. Charpie

and P.W

Mac.4ro!:

Conserving

ewrgy

VI the production

oj‘alutmnum

2. What is known and what has been said of energy conservation

In the past few years a number of organizations and individuals have written reports on potential conservation of energy in the industrial sector related directly or indirectly to the domestic aluminum industry. A number of those reports are important for evaluating conservation policy. They are reviewed here for their major findings-not to provide an exhaustive review, but rather to elicit and evaluate their most widely-held conclusions before going on to synthesize within a model framework the outlook to 1985. An Oak Ridge study [Bravard et al. (1972)] sets the stage as one of the first efforts to account systematically for both direct and indirect energy uses in the production and recycling of various metals. While not directly concerned with an estimate of the conservation potential in aluminum, the calculations show where energy is expended in aluminum production and therefore where Research and Development (R&D) should focus to develop energy-conserving technologies. The conclusions relevant to aluminum production are that there is: (1) High energy content of primary aluminum ingot (no fabrication energy included), and large variation in energy requirements as a result of differing bauxite ore concentrations:” 300; bauxite ore 93,300 BTU/lb.

50% bauxite ore 79,700 BTU/lb.

(2) Energy intensiveness of aluminum ingot produced (clays and shales) other than bauxite ores:12 C/a? 104,800 BTU/lb.

from domestic

sources

Anorthosite 114,600 BTU/lb.

(3) Reduced energy use from recycling. The energy consumed in recycling aluminum is 3-4 T/, of the energy expended in producing ingot from SO’,‘; bauxite ore.13 The data used in this study were quite unrefined, most having been derived from averages quoted or issued in chemical engineering textbooks. A later article by Davenport and Peacey (1974) assesses costs of major alternative technologies for aluminum production that may challenge the current Bayer-Hall process. The Alcoa, Alcan, Monochloride, and Toth processes are discussed. The cost estimates are summarized in table 1. Although it is not demonstrated that the Alcoa process has a distmct cost advantage over Bayer-Hall, Davenport projects that the Alcoa chloride ’ I Bravard et al. 121bid. p. 18.

(1972, p. 15).

R.A. Charpie

and P.W

MacArop.

Conseroing Table

Costs assoctate

with alternative

energy

in the production

of aluminum

1

methods

of alummum

productton.

New Bayer-Hall

New Alcoa I

Alcan

Monochloride

Toth

Capttal cost (S,inet tons/year)

51,545

$1,470

8850

$930

$930

Electricity (kWh/lb.)

6.4

5.1

11.5

6.7

1.0

Carbon (lb./net

450

360

990

765

4.500

15.9

17.5

15.9

13.5

23.9

tons/year)

Direct operating (c/lb.)

15

costs

“Note. scale = 110,000 tons/year.

Source:

Davenport

and Peacey

(1974, p. 28).

process is the strongest competitor to Bayer-Hall.13 Many of the conclusions are based on highly speculative information, but perhaps the best indicators of costs that can be obtained (given the proprietary nature of R&D projects) are found here.14 Moreover it is likely that the forecasts of costs are too low because calculations of direct operating costs have not allowed for the recent rapid escalation in electricity and raw materials prices. Even so, the implications for energy policy are evident -push towards low electricitycarbon usage by taking on high capital-cost processes. The study by Gyftopolous et al. (1974), compares the present energy demand of the aluminum industry with the thermodynamic minimum energy requirement. The results indicate that the present demand is 6.5 times larger than the thermodynamic minimum. This result, coupled with engineering process analyses, leads to the conclusion that large conservation potentials exist to the extent that energy consumption could be reduced 32 percent by 1980.” As such, this is an excellent engineering review but it neglects the economic or cost considerations that determine actual operating modes for aluminum production. What is needed is an economic assessment of the costs of the additional inputs necessary to attain the thermodynamic mimmum of fuel usage. From that finding, the cost of adding capital and reducing energy can be compared and a decision made as to whether it is ‘worth it’ from the point of view of the economy to save energy. The Conference Board study (1974) done for the Energy Policy Project of the Ford Foundation was based on survey data collected from five aluminum compames - Alcoa, Kaiser, Reynolds, Martin Marietta, and Eastlaco. It attempted to create a framework for analyzing the energy conservation 131bid. p. 19. “Davenport and Peacey (1974, p. 28). ‘%yftopolous et al. (1974. pp 69. 77).

26

R.A. Churpie

und P.W MacArov,

Conserrmg

energy

in the productron

ofaluminum

potential and then to consider aluminum in detail as a case study. In a manner similar to the preceding studies, the study breaks energy use in 1980 down by process and concludes that the production of one pound of aluminum ingot will require 8 DOless energy in 1980 than in 1971, because of further adaptation of the energy saving process. However due to increased aluminum capacity total energy use in aluminum is expected to increase from 975.5 x lOI BTU in 1971 to 1.531.4 x 10” BTU in 1980.rh The Report intimates that technology could reduce consumption more. but rt recognizes that new technology requires long lead times for diffusion through the industry. The report stops short, moreover, of doing any formal economic analysis to measure the induced technical progress in energy saving that can be expected to result from increased energy prices. Perhaps the most influential of the reports has been that of the Federal Energy Administration (1974). The work on aluminum is represented in two volumes, the summary volume (Project Indepencience Reporf) and the Task Force Report (Energy Conservation in the Manufacturing Sector 1954-l 990). The Task Force Report provides most of the background information on aluminum, and evaluates existing technologies, future technologies, scrap recycling, and conservation strategies but at a level of sophistication below that of the other studies. The only calculation of interest is one which projects future aluminum energy demands by projecting the energy demand of new and replacement capacity through 1990. Though the Task Force Report can be criticized on grounds of vague generality, the major criticism must be directed at the summary volume. which misrepresents the results of the Task Force document. While neglecting several crucial points raised by the Task Force Report, namely that capital intensity and rigid technological constraints limit the possibility of large, immediate energy savings, the summary sees tit to make statements on four important issues as follows: (1) The conservation potential allows 25 yO reduction in BTU/lb. by 1985.” But nowhere in the Task Force Report can such a figure be found; in fact the Task Force cites a potential savings of 19.6”#; in BTU/lb. by 1985.” (2) The aluminum industry can attain savings of J5-200/, in smelting energ) of input fkom increased efficiency alone. l9 This is based on experience other countries where energy costs have traditionally been higher. This neglects the fact that European smelters use a different form of alumina and a slightly different technology. Therefore the 15-20”” savings could ’ bConference Board (1974. p. 524). “FEA Prqect Independence Report (1974, p. A-162). “FEA Task Force Report (1974, pp. 686). “FEA ProJect Independence Report (1974, p. A-161).

R.A. Churpie

and P.W! MacAvoy. Conseroing energy in the production ofalumrnum

27

only be achieved by large capital investment in new equipment in alumina and smelting plants. In addition, if the U.S. aluminum companies were to operate at this energy-minimizing point, the total cost of manufacturing, aluminum would rise because this is not the most efficient point in relation to other U.S. factor input prices. (3) The U.S. could reduce domestic energy consumption by buying alumina in foreign countries, leaving it to them to put their energy into their aluminu before they sell it to us. ” To carry this argument further, it appears that the U.S. has the opportunity to reduce domestic energy consumption by transferringthe United States aluminum industry to Africa. where production is based on hydropower, having them then export the final product to us. Unfortunately the summary volume does not provide any analysis of the political and economic effects on the U.S. of buying energy in this form rather than as crude oil imports. (4) After suggesting that the U.S. should export energy use in process. the summary volume also concludes that new sources of domestic bauxite may ease our dependence on foreign sources.21 The summary fails to mention, however, that these alternative sources are more expensive and more energy-intensive than the foreign sources.

“‘Ibid, p. A-161. “Ibid, p. A-161. The reports use a variety of energy accountmg conventions; therefore, though energy accountmg results within any report are consistent, it IS difficult to compare forecasts of savmgs among reports. Electrical energy can be converted to heat equivalent m two ways: (1) on the basis of the heat that a kWh wdl deliver (=3.412 BTU/kWh. the net energy accountmg convention) or (2) on the basis of the heat required to produce a kWh (= 10,500 BTU/kWh. the gross energy accounting convention). The gross energy convention IS preferred when analyzmg the energy demand of a single firm or industry because the net energy accountmg convention introduces signihcant biases with respect to purchased electricity In the case of the alummum industry this IS an important problem since a large portion of the electricity purchased now by the alummum Industry is generated from hydropower at a conversion rate near 3.412BTU/kWh. But what counts from a policy standpoint is the energy demand of new and replacement capacity, and smce virtually all of the economic domestic hydro sites have been developed the appropriate rate is that for new Iiscal plants. This IS not to indicate that the net energy accountmg convention and the 7,859BTUikWh conversion factor have no apphcations, but rather that the gross energy accountmg convention is the proper convention to apply when domg policy analyses of the future energy demand of a smgle domestic industry The ORNL Report and the FEA Task Force Report both use conversion factors that attempt to account for the hydropower element of the aluminum energy supply mix: in fact. the FEA Tash Force Report uses both the Gordian Associates factor of 7,859BTLJ/kWh and the net energy accountmg convention of 3,412BTU/kWh. The Gyftopolous Report uses a conversion factor of 10.000 BTUjkWh for all electricity used by the aluminum industry: the Report recogmzes the hydro element but argues that ‘Because the extensive mterconnection of U.S. electric utilities permits the ready exchange of power between regions, alummum production must be regarded as a load on the entire electricity grid and as a dram through that grad on scarce petroleum and gas resources.’ The Conference Board Report uses a conversion factor statement of the rationale for neglectmg the hydro element.

of 10.500

BTUjkWh

without

any

28

R.A. Charpie and P.W MacAvoy, Conserrrng energy in the production of aluminum

Considering all the reports together, the general results to date are quite limited. There has been general agreement that, although energy use per pound of production can be reduced by policy or by likely autonomous gains in efficiency, the total amount of energy used in this industry will increase substantially in the coming decade. But there has been generally (1) failure to provide information on the costs of implementing various conservation policies, (2) failure to specify the data used in the analysis, (3) failure to consider methodologies other than engineering process analysis as a means of projecting future energy demands, (4) failure to use the correct set of energy accounting conventions.22

3. Improving the estimates

of reduced energy consumption

The forecast of energy savings may be improved by using a variety of methods to estimate each of three types of future energy savings, ‘one time’ reductions. incremental reductions due to readjustments in response to rising energy prices, and long term reductions due to technological change. 3.1. Non-recurring

snLiings

The non-repeatable savings are best estimated by an interview analysis similar to that used by the FEA to estimate total energy savings. Discussions with aluminum plant managers should yield an estimate of the ‘kink’ that increased awareness of the need for efficiency in energy usage will yield. Among the experts in this field of housekeeping operations is Dow Chemical Company. Dow markets a consulting service that shows other companies how to save energy in their manufacturing operations. The Dow experience has shown that savings on the order of 25% may be anticipated in a typical manufacturing plant following the initiation of an aggressive energy accounting and use-reduction program, although individual plant savings may vary from 109h to 5Oq,. This estimate must be considered- to be too high for an aluminum plant since the major energy usage is in smelting, and this process is a tine-tuned, rigid one that does not lend itself to process modifications that vary energyother inputs combinations. Some insight into the savings achieved in recent years by aluminum companies may be gained by examining some public remarks by corporate representatives: (1) In testimony California.‘”

before Richard

the Public Utilities Commission Pool of Kaiser Aluminum showed

of the State of that 11 I’(, of the

“Gyftopolous et al. (1974. p 76) ‘?estlmony of Richard B. Pool, Kaiser Aluminum and Chemical Corporation, before the Publx Utlhtces CornmissIon of the State of Califorma regardmg application no. 54279 of Pacific Gas and Electric Co.. for an increase m electrlclty rates.

R.A. Churpie and P.W MacAvoy. Consewing

energs tn the production of aluminum

29

total electricity usage in Kaiser’s California fabricating plants has been saved. Pool noted, however, that this auxiliary electricity usage is a small portion of the total energy usage in aluminum manufacture. (2) In a speech at the Conference Board, Mr. A.C. Sheldon of Alcoa compared the energy demand for a variety of mill products in 1973 with the energy demand in 1971. He cited percentage reductions on the order of 300,; in mill products. Sheldon also noted other savings including an improvement in furnace thermal efficiency of 61 yh via preheating of charges and combustion air, a 50/0 savings due to reduction in lighting loads, and average reductions in plant space heating of 11 pi,. (3) At the MIT Aluminum Energy Conference, Mr. James Cole of RJR Archer stated that energy conservation programs at RJR Archer’s foil mill have resulted in savings of approximately 13 “/ of the total energy usage.24 No one percentage can adequately encompass the variety of one-shot savings. The major segment of the manufacturing sequence where one time savings are obtainable is the fabrication level of the industry, where the 257; savings proposed by Dow Chemical (and echoed in the Alcoa tigures) seems to be a sound estimate. Negligible housekeeping savings are anticipated in smelting with somewhat larger one-shot savings in alumina refining (approximately 5% of the total energy usage in the Bayer plant) where the primary focus appears to be on steam systems.2s 3.2. Incremental

changes

As prices for energy increase, adjustments will be made in processes and investment to ‘save’ on the more expensive energy inputs. By how much? A useful methodology for estimation is the extrapolation of historical trends of energy usage in the aluminum industry. Careful examination of historical data in the Census of Manufactures has led to the conclusions that large portions of these data are suspect and that the only consistent set of numbers appears to be that on electricity use in smelting, as follows: 194012 kWh/lb.; 1954-9 kWh/lb.; 1958-8 kWh/lb.; 1962-8.6 kWh/lb.; 1967-8.2 kWh/lb.; 1971-8.1 kWh/lb. All these figures are in direct current kWh/lb. The observed rate of decline might lead one to conclude that the industry is approaching a limit on kWh reduction in the aluminum production process. Thermodynamic calculations, however, show that the theoretical minimum yields a value of 3.7 kWh(t)/lb. Al allowing for the fact that at present the energy for smelting is supplied partially as electricity and partly as carbon in the form of electrodes. The stoichiometry of the Z4Flemmgs et al. (1974, p. 20). “Conference Board (1974, p. 549). a

30

R.A. Charpie and P.W MacAvoy, Conservmg energy m the producfion of aluminum

reaction: AllO + 3/2 C =2Al+ 3/2 CO? requires a minimum of 0.33 lb. of carbon per pound of aluminum. or 1.37 kWh(t)/lb. Al of available useful work in carbon form. Adding this to the minimum amount of electricity 2.33 kWh(e)/lb. Al the total practical minimum is 3.7 kWh(e)/lb. Alz6 On this basis the conclusion is that the industry has not yet approached the practical limits of the reduction process (even though this calculation does not allow for the complex heat balance requirements that restrict the operating range of a potline). In fact, the ‘limit’ probably lies between 8.1 and 3.7 kWh, and methods should be developed to estimate it more closely. A model of aluminum energy demand could be specified and estimated using historical data that predicted this limit. Using a cost formulation, one might regress energy use per unit of output on some index of energy prices relative to labor. capital, and raw materials costs. Equivalently, using the dual to the cost formulation, one could start with some assumptions about aluminum technology and attempt to estimate the aluminum production function. If energy. labor, and capital were substitutes for one another, one would anticipate that increases in the relative price of energy would bring about reductions in energy demand through induced technical progress aimed at substituting relatively less expensive capital and labor inputs for energy inputs. Conversely, increases in the relative prices of capital and labor would increase the energy-intensiveness of aluminum manufacture.” The practice would be to observe such changes in the input mix using both time series and cross-sectional data. The use of cross-sectional data would be based on the premise that input prices vary between different regions of the country, so that at any point in time one observes different factor input mixes in different regions of the country m response to these different prices. For example, a comparison of aluminum technologies in the Tennessee Valley Authority (TVA) and the Bonneville Power Administration (BPA) regions would prove interesting. These two regions are prime locations for alummum smelting plants because of the availability of large quantities of inexpensive electricity. but in the BPA region electricity sells for roughly l/2 to l/3 of the TVA price electricity price. This price differential is due to almost all of the BPA’s electricity being generated by hydropower, while the TVA has exhausted its hydropower potential and supplies a large portion of its electrical output from coal-fitted steam generating plants. One would expect that a comparison of TVA plants and BPA plants would reveal a more energy-intensive technology in the BPA region. Unfortunately regionally disaggregated data detailing energy demand and ingot output were not available for testing this hypothesis. ‘6Gyftopolous et al. (1974, p. 69). “For an example of technological progress almed at reducmg the demand expenwe factor Input. see the work on the petroleum Industry by Enos (1962).

for the most

R.A. Charpie and P.W. MacAvoy, Conservmg energy in the production ofalummum

31

The use of time-series data might reveal the same change in factor mix as relative prices changed over time, rather than across regions. The Census data cannot be used to confirm this expectation-energy demand per unit has decreased during periods of relatively decreasing energy prices. This pattern may be explained in a number of ways. The six nationally aggregated data points are too sparse to permit any statistically sound conclusions; their interpretation may be confounded due to aggregation errors resulting from the pooling of data from different geographical regions. Another explanation, in accordance with the prevailing industry wisdom, is that reduced energy demand per unit output is the result of an historical learning curve process based ultimately on slowly improving technology. Such an effect might be specified by introducing ‘time’ as an independent variable, as a proxy for technological change, into a model with energy consumption as the dependent variable. Alternatively the effect would be introduced into the model by formulating the equation to state that energy is not a substitute for capital and labor but is a complement to one or the other of these inputs. In this interpretation, energy usage reductions would correspond to technical progress aimed at reducing the use of capital and/or labor inputs.28 At present, however, any effort at this type of econometric modelling is once again limited by the lack of reliable data on both dependent and independent variables. When and if such data do become available they may be used as guides to the proper specification and estimation of a model of aluminum energy demand which may then be used as a forecasting device. All that can be done at this stage is to examine the manufacturing process sequence, and to collect engineering estimates of likely future technical progress at each stage in the sequence. This process analysis is typical of the work done in attempting engineering estimates of production functions. Information is collected by experiment and from the engineers’ experience in the daily operation of a technical process; this ‘productivity’ information is then combined with cost considerations to estimate the path that technological progress will follow. 4. Process analysis estimates

of future incremental

energy reductions

The critical process in aluminum manufacturing is the Hall cell, and no major changes have taken place in the Hall process since its implementation at the turn of the century. The historical decline in kWh/lb. energy usage is the result of tine tuning potlines over time. The present industry average usage is 8 kWh/lb., while the best existing commercial technology requires 6.5 kWh/lb. One conservation measure which might be expected in the future will be the diffusion of this 6.5 kWh/lb. technology throughout the industry, though this will require large capital expenditures to overcome back reaction of aluminum and resistance in the cell flux and hardware. “For

a discusslon

of the complementarlty

of energy

and capital,

see Berndt

and Wood

(1974).

32

R.A. Charpie and P.W MacAvoy. Conserving energy in the production ofalummum

Such back reaction of aluminum to alumina is endemic to the process and is the result of molten aluminum reacting with oxygen or carbon dioxide formed at the anode and diffused through the flux bath. System resistance accounts for 657, of the voltage drop across the Hall cell, most of which (90%) is attributable to the resistance of the anode, the cathode, and the conductors and the residual of which (35 %) is used for the reduction process and the heating of the chemical inputs. Incremental improvements in the Hall cell depend upon alleviating these two technical constraints. Back reaction may be reduced by increasing the spacing between the anode and the cathode; however, this has the undesirable effect of increasing the system resistance and the resultant electricity demand. Conversely, the system resistance may be reduced by decreasing the anodecathode distance, but this will increase the back reaction. Specific modifications to the Hall cell include: (1) Reduction of current in the Hall cell by 15 % to reduce resistance losses; this would result in electrical energy savings of 15 “/‘,, but will also result in reduced production per cell. (2) Lowering of electrolyte resistance by modifying electrolytes composition to improve conductivity. (3) Reduction of anode resistance by developing new anode materials. (4) Reduction of resistance by reducing distance between anode and cathode; as explained this is limited by the degree of back reaction which can be tolerated -in addition reduced interpolar spacing can result in shorting of the pot. (5) Computer control of the heat balance in the pot so that the system operates slightly above the flux melting point; this will reduce the turbulence of the bath at the metal-flux interface where back reaction occurs. Each of these modifications requires the solution to difficult engineering problems due to the delicate heat balance in the Hall cell and the interplay between system resistance and back reaction. It would be useful to estimate the costs of each of these potential modifications to determine which changes will be adopted solely in response to rising energy prices. Cost data could be developed however only for current reduction, the first modification. Smelting pots can be operated across a range from (1) the point of maximum output per unit time and (2) the point of minimum energy use per unit time. A comparison of these two operating modes, one a high current, high output mode (160 kilowatt ampere (KA) into the pot) and the other a low current, low output mode (135 KA into the pot), is shown in table 2. The calculation shows that the price of delivered electricity must exceed 12.1 mills/kWh before operation of a high anode current density type cell is shown to be economically inferior to operation of a low anode current

R.A. Charpie and P.W MacAvoy, Conserving energy in the production of alummum

33

Table 2 A cost compartson Capital

of two alternative

aluminum

plants

usmg different

current

KA

160 -

cost

8,960 4.370 9,959 24,581 54.577 15.017 11.894 1,045

Site development Servtce buildings Materials handhng Carbon plant Pot rooms Power supply system Foundry Mobile equipment

130.403 40,425

dtrect cost (BDCC- 1973) mdtrect costs & owner’s costs fj 313: total (1973 startup) Ftxed charges Operatmg

ia 15 “& per year.

S/MT metal

-135 KA 8.960 4.370 9,959 24.58 1 62.764 17.270 11.894 1.045 140,843 43.66 1 -.-

MSi170.828

184,504

158.174

170.837

equal

cost

equal 18.219

Raw materials Labor & supervision Power supply c@,S.012 per kWh 2.819.815.740 kWh 2,475,822,510 kWh Property taxes 3 2 % x capttalization Other costs

densities.’

20.041

33,838 3,400 equal

index, annual $/MT metal

MS 55,457 342.33

29.7 10 3,680 equal 53,631 331.06

Index Operattons

- expenses

& fixed charges

%/MTmetal breakeven power

balance, %/MTmetal prtce 5.0121 per kWh

500.50

501.89 +1.39

‘Nore The assumpttons are as follows: (1) These plants are presumed to have been built in 1973, and to operate at a capactty of 162,000 metrtc tons (MT) per year. (2) In reducing amperage density by lSo/, (160KA to 135KA) the capital cost of pot rooms and power supply the 162,000 annual metric ton capacity. (3) The power system mcreases by 15”, to mamtam eflictency at the current denstties is at 160 KA -7,894 alternating current kWh/lb. metal. total demand. and at 135 KA - 6,631 alternating current kWh/lb. metal, total demand.

density type electrolytic cell. In general, for an older smelter the use of a relatively inefficient power system is financially rewarding, even in the face of power costs exceeding 12.1 mills/kWh. Only when the average total cost of a new smelter using low current densities would be less than the average variable cost of an old smelter using high current densities, should the old smelter be scrapped in favor of the new smelter.29 With present electricity costs well below 12 mills/kWh, aluminum companies operate closer to the point of maximum output per unit time because this has proven to be closer ‘“See F‘ellner (1951).

34

R.A. Churpre and P.W MucAvo.~, Consercmg

energy in the producrlon

o/aluminum

to the point of minimum total cost for production. Generally, the cost function of the aluminum industry is dominated by the large capital outlays necessary to build a plant close to efficient scale. The operating costs associated with electricity are dwarfed by these capital costs; therefore. operation at the energy minimizing point would force an increase in total costs to the industry. In addition to potential changes in the Hall cell, a number of changes in anode systems could influence energy consumption levels. Two basic systems are in commercial use: Soderberg and prebaked. In both cases the anode raw material is a paste of petroleum and low-ash petroleum coke. The Soderberg technique feeds the paste continuously into the electrolysis pot, where the heat of the smelting pot bakes out the volatiles. The prebake system uses natural gas to bake anodes prior to their use in a smelter. Anode baking energy consumption is lower for prebake systems (1,130 BTU/lb.) than for Soderberg systems (5,700 BTU/lb. A1).30 In 1971, 33 Y0 of the smelting capacity used Soderberg systems while the remainder used prebake anodes.31 In the future it is anticipated that Soderberg systems will not be built in new facilities and may be replaced in existing plants. This is not due to reduced energy intensiveness of the prebake system but is a result of environmental regulations controlling the release of volatiles. The prebake system permits better control of volatiles than the Soderberg system. A switch from gas to oil in the prebake energy mix is predicted, but this will not have much effect on energy usage. As indicated in the Davenport and Peacey article, there are some radical technical changes on the horizon which may sharply influence the energy demand for smelting. 32 The first of these processes is the Alcoa chloride process. A current experimental plant of 15,000 tons/year capacity is under construction in Palestine, Texas. This capacity will be expanded to 300,000 tons/year by 1981 making it larger than any currently existing U.S. plant. The Alcoa process starts with Bayer processed alumina which is chlorinated under reducing conditions to produce gaseous aluminum chloride. This gas is condensed to solid form and fed continuously into an electrolytic cell where it is dissociated by direct current into liqmd aluminum at the cathode and gaseous chlorine at the anode. This chlorine is then recycled, so the process operates as a closed system. This process involves reduced electrical consumption over the Hall cell due to increased electrical conductivity and reduced inter-polar spacing. Alcoa claims that this process will yield an state-of-the-art technology, electric energy saving of 30% over present resulting in an energy requirement of 4.5 kWh/lb. A second process of interest is the Alcan process. Alcan developed this ‘“Conference Board (1974. p. 533). “Ibid. p. 533 3’The techmcal details are drawn from Davenport

and Peacey

(1974)

R.A. CharpIe and P.W MacAvoy, Conserving

energy

in the production of aluminum

35

process and built a 5,000 tons/year pilot plant that was later shut down due to stress corrosion problems. Alcan says that these problems have been solved but no work has gone forward on the plant since 1968. The benefits of the Alcan process are reduced capital costs and the ability to operate at a smaller scale; these are counter-balanced by the increased direct electricity consumption of the Alcan process (11.5 kWh/lb.) and the increase in carbon consumption of 1207; over the Hall reduction cell. A third process is the Monochloride process which has been demonstrated only on a laboratory basis. This process produces aluminum directly from bauxite, which eliminates the Bayer processing step. The Monochloride process exceeds the present Bayer-Hall technology in terms of direct electricity requirement (6.7 kWh/lb.) and carbon costs (70% greater) but compensates by offering sharply reduced capital costs. A fourth process is the Toth process which has caused some controversies among engineers with respect to its feasibility. The process is based on the exchange reaction between manganese metal and aluminum chloride to give, manganese dichloride and aluminum metal. The use of the manganese metal as the reductant in the Toth process in the view of many renders the process economically infeasible. The status of each of these challengers to the Bayer-Hall technology is summarized in the cost table below. This table augments the results of the Davenport article by: (1) allocating a fixed cost to each pound of ingot using an 18 percent cost of capital for aluminum companies, and (2) introducing a column headed ‘New Alcoa II’ which represents the judgment of an aluminum industry financial analyst regarding Alcoa’s cost parameters for Table 3 Projected

total costs assoctated

Capital cost (J/net tons/year) Fixed costs (e/lb.) Electrictty (k Wh/lb.) Carbon (Ib.;net tonsyear) Dtrect operatmg costs (c/lb.) Total costs (e/lb.)

wtth alternattve

methods

of alumtnum

productton.”

New Bayer-Hall

New Alcoa I

New Alcoa II

Alcan

Monochloride

Toth

$1.545

s 1,470

%l,OOO

$850

9930

6930

139

13.2

9.0

7.6

84

8.4

64

5.1

4.5

11.5

6.7

1.0

450

360

360

990

765

4,500

15.9

17.5

13.9

15.9

13.5

23.9

29.8

30.7

22.9

23.5

21.9

32.3

“Note The scale of plant chosen for analysis is 110,000 short tons/year, mformatton was avatlable on present and potenttal cost at that level of operations.

smce

more

36

R.A. Charpie and P.Wr MacAcoy, Consercmg energy m the production of aluminum

the chloride process. This table reveals sharp disagreement between the results of the Davenport article and the judgment of the industry analyst: Davenport projects the Alcoa process as slightly more expensive than the Bayer-Hall process, while the analyst shows them to be much lower. But these cost figures suffer from the same problems that are mentioned with respect to the original Davenport article-operating costs are probably underestimated because they were calculated assuming 6 mills,‘kWh electricity prices, and they do not anticipate recent bauxite price increases. The only process which can be considered as a replacement for the BayerHall process in the near future is the Alcoa chloride process which is on the way to full-scale commercialization. The other processes can be eliminated on the following grounds: (1) the Alcan is not likely due to higher power requirements and technical obstacles. (2) The Toth process is unattractive due to high operating costs. (3) The Monochloride technique is still in the research stage and cannot yet be considered a serious alternative. Energy reductions can also be achieved by increased scrap recycling. The *energy input required to bring recycled scrap from the plant door to a useful molten state is about 6,000 BTU/lb.,33 i.e., 7S; of the energy required to make an ingot from raw materials. There are obvious limits, however- the system cannot recover more than a small percentage of previous production because the collection costs are too great. There are two kinds of scrap, that which is generated by plants making end products, and old scrap which is recovered from metal that has been used by consumers. In 1971 both old and new scrap accounted for 21 o/0 of total domestic primary plus secondary production. From an energy conservation standpoint, the companies could try to increase the old scrap recycling rate above the present 4”,; value, while decreasing or at least holding constant the 17 “; new scrap rate by improving the efficiency of fabrication operations. Calculations have shown that the old scrap recycling rate could be increased by expanding investment in recycling systems so that recycling provides a maximum of 33.5% of production or equivalently that only 0.25 x lo6 tons of old scrap could be recycled out of a total potential old scrap volume of 1.8 x lo6 tons, resulting in 1.55 x IO6 tons of old scrap being discarded.34 Then the question is how much will recycling increase by 1980. The answer depends in part on why the present aluminum old scrap recycling rate is so low. This is the result of many factors, including ignorance and resultant waste on the part of consumers in handling recycling materials. In addition, certain practices may have limited the growth of the secondary industry, such as zoning and licensing restrictions. depletion allowances enjoyed by the primary producers, and discrimmatory transportation freight rates. The freight rate pattern for aluminum is similar to steel: “3Gyftopolous et al. (1974, p. 72). “Flemings et al. (1974, p. 2).

R.A

Charpie

and P.W

MacAm),

Consemng

energy

III the producrrm

qf‘alum~num

37

it is 50 ‘,i more expensive to move an equal number of useful iron units in the form of scrap than as iron ore. 35 This effect retards the present scrap market by encouraging usage of ore and affects future scrap markets by diverting new investment from scrap-intensive to ore-intensive metal making. There is an additional reason. The secondary industry is geared to the production of castings, while the best opportunities for increased recycling is the aluminum can. It is dillicult to recycle cans into castings because of alloying mismatches. In all at present there is no way to assign weight to these factors, and extensive data collection and analysis is necessary before judgments can be made in order to predict future rates of recycling. Given all these factors and conditions, how much will energy consumption change in the near future? One can attempt to predict this change by combining the results of the previous studies as parameters and values of variables in a crude aluminum energy simulation model. The model replicates the major steps in aluminum manufacture (mining. Bayer-Hall, fabrication) to project the primary aluminum energy production in any given year, and then combines this with a projection of secondary production to arrive at a calculation of total output. Using input-output coefficients, the average energy demand per pound of aluminum in that year is calculated and then using the total production estimates the total energy consumption is calculated. The model operates under the following assumptions: (1) Existing 1971 capacity has a direct energy demand of 7.8 kWh/lb. (where ‘existing’ is defined as capacity that was built before or during 1971). (2) All anode capacity built after 1971 uses prebaked anodes. (3) One-shot savings in mining are 5 ‘4 of base year 1971 consumption. (4) One-shot savings in Bayer are 5 V,; of base year 1971 consumption as well. (5) There are one-shot savings in smelting. (6) One-shot savings in fabrication are 257; of base year 1971 consumption. (7) Ancillary electricity usage remains at 0.6 kWh/lb. for all potlines. (8) The base year energy demand for aluminum is 128.500 BTU/lb. These assumptions are derived by combining the results of the studies referred to previously. But they also follow from the appraisal of all available sources made in the previous sections, as expanded on by interviews and conversations with industry and government experts on energy utilization. The simulation model itself is composed of ten equations. A listing of the equations and the associated variables are shown in table 4. To use this framework for making projections, one must supply projections of secondary aluminum capacity; the direct current kWh/lb. electricity demand of replacement; new and Alcoa chloride primary capacity; the energy required for aluminum recycling in BTU/lb. A variety of simulations have been carried out using this accounting model, the first denoted as ‘FEA Scenario’ 35Barnes (1972)

38

R.A. Chorpie and P.tl! MaeAttoy, Consertling energy m the production ~~ulurnitlurn

since it is based on the following set of parameters suggested by the FEA Task Force Report: (1) The capacity projections are taken from projections of economic growth, assumed by FEA, that stipulate that non-ferrous metals will grow at approximately 4.5% per year through 1990. (2) The FEA has assumed that existing capacity is replaced at the rate of 1.57; per year, and that all new capacity after 1980 wiI1 be Alcoa chloride capacity.36 (3) New capacity energy demand is 6.5 kWh/lb. (4) Replacement capacity energy demand is 6.8 kWh/lb. (5) Alcoa chloride capacity energy demand is 4.5 kWh/lb. (6) Recycling energy demand is 6,000 BTU/lb. These simulations were compiled for 1975, 1977, 1980, 1985 and 1990, as presented in table 5. A second set of simulations was attempted in order to analyze the sensitivity of the results to the successful diffusion of the Alcoa chloride process. These simulations used the same parameters as the FEA simulations 36FEA Task Force

Report

(1974, pp. 6-84).

Table 4 Equations

m the aLcountmg

model for energy

utlhzation.

Purpose

Equation

one-shot

savings

Determines the one-shot m Bayer processing

savings

D.FAB = P.FAB * 20,000

Determines the m fabricatmn

savings

PRIMECAP

Determines the total primary pa&y in any given year

D.MINE

= P.MINE

Determmes in mmmg

* 2,000

D.BAYER=P.BAYER

* 12,000

= EXCAP + NEWCAP -t RECAP + ALCOACAP

the

one-shot

ca-

DSMELT=91,7OC-(8.4*EXCAP +(NEWKWH +0.6 * NEWCAP + (REPKWH + 0.6 * REPCAP (ALCOAKWH +0.6) l ALCOACAP) * 1.04 * CONV/PRElMCAP

Determines the savings m smeltmg given the projected capacities and energy demands of the various types of alummum smelting capacity

D.ANODES

Determines the savings m anode energy consumption given that all new capacity uses prebaked anodes Adds the results of the previous two operattons to determme the total savmgs m the Hall process

D.HALL

= 2.800 - (5,600 + EXCAPI3 + 1,200 * (0.667 + EXCAP -REPCAP+NEWCAP),‘PRIMECAP

= D.ANODES

D.DEMAND

=D.MINE

+ D.SMELT

+ D.BAYER

+ D.FAB

Determmes the total energy mgs for primary aluminum Determines the primary energy demand

DEMAND

= (PRIMECAP t PDEMAND * SDEMAND)/(PRIMECAP

+SECONDCAP + SECONDCAP)

sav-

aluminum

Determines the energy demand of tbe alummum Industry given the relative magnitudes of the primary and secondary capaclttes

R.A. Charpie

and P.W! MacAvoy.

Conserving

energ_r in the productton

39

of &mm~m

Table 4 (continued) Vartable

names

Delinttton

(units)

ALCOACAP

Alcoa chlortde pacrty (lb./year)

ALCOAKWH

Efectrtcrty demand of Alcoa chlortde capacity (direct current kWh,‘lb.)

CONV

Gross

energy

productton

ca-

conversron

factor

( = 10.500 BTU;kWh j D.ANODES

Change m anode (BTU/lb.)

D.BAYER

Change m Bayer process demand (BTUilb.)

D.DEMAND

Change m prtmary alummum ergy demand (BTU,llb.)

en-

D.FAB

Change mand

de-

D.HALL

Change in total (BTU/lb.)

D.MINE

Change in mining (BTU/lb.)

DSMELT

Change mand

DEMAND

Energy demand dustry (BTU)

EXCAP

Existing Bayer-Hall capactty (lb./year)

NEWKWH

Electncity demand Hall productton current kWh/lb.)

P.BAYER

Percentage process

P.FAB

Percentage

savings

in fabrication

P.MINE

Percentage

savmgs

in minmg

PDEMAND

Primary alummum mand (BTU/lb.) Primary aluminum (lb./year)

PRIMECAP

energy

m fabricatton (BTU/lb.)

demand energy

energy

Hall cell demand energy

in smeltmg (BTU/lb.)

demand

energy

de-

of aluminum

m-

production of new Bayercapactty (direct

savings

m

Bayer

energy

de-

capacity

REPCAP

Replacement Bayer-Hall inum capacity (lb./year)

REPKWH

Electricity demand of replacement Bayer-Hall alummum capactty (direct current kWh/lb.)

SECONDCAP

Secondary capacity

SDEMAND

Secondary alummum mand (BTU/lb.)

aluminum (lb./year)

alum-

productton energy

de-

Table 5 FEA scenarios.

Year

Energy demand

1971

101,900 BTU/lb.

1973 1977 1980 1985 1990

94,400 BTU/lb. 92,800 BTU/lb. 90.600 BTU/lb. 84,400 BTU/lb. 79,600 BTU/lb.

O0Reducrlon m energy demand (197 1 base) -.

-

7.4 *,0 8.9 :b Il.1 9” 11.21 21.9::

8.3 x 1.1 x 1.7 x 3.2 x 5.0 x

lo’* 10” 10” 10’ ’ IO”

BTU BTU BTU BTU BTU

Table 6 No Alcoa chloride scenarto. ‘IO Increase relatwe Year

Energy demand

to FEA scenarto

1971 1975 1977 1980 198.5 1990

101,900 BTU/lb. 94.400 BTU/lb. 92,900 BTU/lb. 9 1.200 BTU/lb. 38,200 BTU/lb. 86,000 BTU/lb.

4.3 “; 8.0“,

Table 7 Afternative recycling scenartos.

recyclmg rate

Energy demand

“b Savmgs relative to FEA scenario

0””

i-70/:,

5”” IOU” 15”” 20U,

22 “f, 21”/, 32 “,,, 37%

97,300 BTU/lb. 91,800 BTU/lb. 86,300 BTU/lb. 80.800 BTU/lb. 75.300 BTU/lb.

-7.40, -1.3”, 4.7 1” 10.8 7,: 16.9:~

scrap recycling rate Old

Total scrap

except that no Alcoa chloride capacity was presumed; instead all of FEA’s Afcoa chloride capacity was re-classified as new Hall capacity. These results, presented in table 6, indicate that energy demand through 1985 is relatively insensitive to the introduction of technological change. A third set of simulations attempted to analyze the sensitivity of the results to the assumed scrap recycling rate. The FEA 1980 parameters were used except that (1) the new scrap recycling rate was maintained at 17 “,& while (2) the old scrap recycling rate was allowed to vary from 0:; to 2090 (the FEA stmulation uses a total recycling rate of 23:~). The results, presented in table 7, show that energy demand is sensitive to the presumed recycling rate. A fourth set of simulations attempted to analyze the effect that employing alternative domestic sources of bauxite would have on energy demand. The

R.A. Charpie and P.W! MacAco).

Conseroing

energy in the productron

of alummum

41

Table 8 Alternatwe

domestic

sources

scenario.

3, of alternatwe domestic sources of bauxite

Energy

demand

a0 Increase relative to FEA scenario

10;; 20 “/O 30 7, 50 ?,

92.900 95,200 97,500 100,000

BTU/lb. BTU/lb. BTU/lb. BTU/lb.

2.5 “,, 5.1:; 7.6 “a 10.4 “;

FEA 1980 parameters were used except that the percentage of alumina derived from alternative domestic sources was varied from 0% to 50 76, where the alternative domestic sources were assumed to have an energy demand 2.5 times as large as existing Bayer plants. The results, presented in table 8, show the ‘price’ (in terms of increased energy demands) that the nation must be willing to pay to reduce our dependence on foreign sources of bauxite. These results are obviously preliminary and require extensions in numerous directions. Specifically some simple econometrics should be done in an effort to model the demand for energy in aluminum as a function of economic variables including the price of aluminum and the energy, capital, labor, and raw material factor prices. Ceteris paribus, one would expect to observe large variations in energy demand per pound across time and across geographical regions due to differences in relative energy prices across these dimensions. At present such an attempt is thwarted by an inability to obtain reliable regional production statistics and regionally disaggregated historical energy data. A second major extension would make alternative capacity projections in the simulation model to determine the range of values that can be anticipated from errors in projections. Capacity is used as a proxy for demand in this simulation model. In fact this full capacity assumption is a poor one because historically aluminum production as a percentage of capacity has fluctuated from 537; to 1059,. In periods of low demand old capacity is shut down first, which will bring a reduction in BTU/lb. of aluminum produced, while in periods of strong demand old as well as new capacity operates at 1009; which will result in an increase in BTU/lb. of aluminum produced. No allowance is made for such selective capacity shut-downs in this simulation model; if demand forecasts were coupled with capacity projections some progress could be made in this direction. Third. better information on one-shot savings should be factored into the model, as soon as such information becomes available. The authors suspect that though 25 “/o one-shot savings may be obtainable in fabrication operations in the long run (after 1985), these changes may not be implemented by 1977 or 1980 and in this sense the model may over-estimate the frabrication

savings in the short term. In addition, the Bayer one-shot savings are poorly documented, and seem conservative. Equally important, more information should be gathered regarding the recycling industry. Research should be initiated to investigate the validity of the projections made here. Detailed data analysis should reveal if and where the growth of the recycling industry is affected adversely by government zoning and transportation rate setting. These issues must be sorted out because they are strong determinants of the future growth of the recycling industry, which in turn strongly influences the energy demand of aluminum. Last of all, more data should be collected regarding alternative domestic sources of bauxite. A survey of the domestic sources to be tapped and a forecast of import prices must be made in order to forecast energy demand more accurately, since that demand depends on the rate of domestic sources. Even though more and better research should be conducted on energy conservation in aluminum, the results indicated here make two points. First, the rate of energy savings per pound of Al production derived from market responses to more expensive energy would not be substantial-at least not so substantial as to significantly affect total energy consumption in this industry. Second, the promising sources of initiative towards larger savings are not the much-discussed new technologies but rather an expanded recycling industry. Public policy here could have substantial positive effects on energy savings by moving towards removing various barriers to the low cost collection and organization of aluminum scrap. References Alummum Association, 1972, Alummum stattstlcal revtew (New York). Barnes, T.M., 1972,The Impact of railroad freight rates on the recyclmg of ferrous scrap, Report to the Institute of Scrap Iron & Steel. Jan. (Battelie Laboratories, Cotumbus. OH). Berndt, E. and D. Wood, 1974, Technology, prices. and the dewed demand for energy (MIT, CambrIdge, MA). Bravard, J.C., H.B. Flora and C. Portal, 1972, Energy expendttures associated wtth the productjon and recycle of metals, Oak Ridge National Laboratories Report no. ORNL-NSFEP-24. Nov. (Oak Ridge, TN). Charpie, Richard A., 1975, The energy content of high performance super-alloys-A technical and economic analysis. Master’s Thesis (MIT, Cambrtdge, MA). Conference Board, 1974, Energy consumption m manufacturmg (Balhnger, Cambridge, MA). Davenport, W.G. and J.G. Peacey, Evaiuation of alternattve methods of alummum productIon, Journal of Metals, July. Enos. J., 1962, Petroleum progress and profits (MIT Press, Cambridge, MA). FEA (Federal Energy Administration), 1974, Project mdependence report, Nov. (Washington, DC ). FEA Task Force on Industrial Conservation, 1974, Energy conservation in the manuracturing sector (1954-1990) (Washington, DC). Fellner, Wilham, 1951, The influence of market structure on technological progress, Quarterly Journal of Economics LXV. Flemings, Merton C., Kenneth B. Higble and Donald J McPherson, 1974, Report of the conference on energy conservation and recycling m the afuminum industry (Cambridge, MA). Gyftopolous, Elias P., J. Lazaros, J. Lazarldls and Thomas F Wldmer, 1974, Potential fuel effectiveness m Industry (Ballinger, Cambridge, MA).