Chapter 4 â€” Fuel Cycles - MIT - Massachusetts Institute of

29C h a p t e r 4 — F u e l C y c l e s

The description of a possible global growth sce-nario for nuclear power with 1000 or so GWedeployed worldwide must begin with somespecification of the nuclear fuel cycles that willbe in operation. The nuclear fuel cycle refers toall activities that occur in the production ofnuclear energy.

It is important to emphasize that producingnuclear energy requires more than a nuclearreactor steam supply system and the associatedturbine-generator equipment required to pro-duce electricity from the heat created bynuclear fission. The process includes ore min-ing, enrichment, fuel fabrication, waste man-agement and disposal, and finally decontami-nation and decommissioning of facilities. Allsteps in the process must be specified, becauseeach involves different technical, economic,safety, and environmental consequences. A vastnumber of different fuel cycles appear in the lit-erature,1 and many have been utilized to onedegree or another. We review the operatingcharacteristics of a number of these fuel cycles,summarized in Appendix 4.

In this report, our concern is not with thedescription of the technical details of each fuelcycle. Rather, we stress the importance of align-ing the different fuel cycle options with theglobal growth scenario criteria that we havespecified in the last section: cost, safety, non-proliferation, and waste. This is by no means aneasy task, because objective quantitative meas-ures are not obvious, there are great uncertain-ties, and it is difficult to harmonize technicaland institutional features. Moreover, differentfuel cycles will meet the four different objec-tives differently, and therefore the selection of

one over the other will inevitably be a matter ofjudgment. All too often, advocates of a particu-lar reactor type or fuel cycle are selective inemphasizing criteria that have led them to pro-pose a particular candidate. We believe thatdetailed and thorough analysis is needed toproperly evaluate the many fuel cycle alterna-tives.

We do not believe that a new technical configu-ration exists that meets all the criteria we haveset forth, e.g. there is not a technical ‘silver bul-let’ that will satisfy each of the criteria.Accordingly, the choice of the best technicalpath requires a judgment balancing the charac-teristics of a particular fuel cycle against howwell it meets the criteria we have adopted.

Our analysis separates fuel cycles into two classes:“open” and “closed.” In the open or once-through fuel cycle, the spent fuel dischargedfrom the reactor is treated as waste. See Figure4.1. In the closed fuel cycle today, the spent fueldischarged from the reactor is reprocessed, andthe products are partitioned into uranium (U)and plutonium (Pu) suitable for fabricationinto oxide fuel or mixed oxide fuel (MOX) forrecycle back into a reactor. See Figure 4.2. Therest of the spent fuel is treated as high-levelwaste (HLW). In the future, closed fuel cyclescould include use of a dedicated reactor thatwould be used to transmute selected isotopesthat have been separated from spent fuel. SeeFigure 4.3. The dedicated reactor also may beused as a breeder to produce new fissile fuel byneutron absorption at a rate that exceeds theconsumption of fissile fuel by the neutron chainreaction.2 In such fuel cycles the waste streamwill contain less actinides,3 which will signifi-

Chapter 4 — Fuel Cycles

MIT_ch04_29-36.qxd 7/16/2003 1:26 PM Page 29

30 M I T S T U D Y O N T H E F U T U R E O F N U C L E A R P O W E R

Natural uranium306,000 MTU/year

Fresh UOX29,864 MTHM/year

Spent UOX Fuel29,864 MTHM/year

Conversion, Enrichment, and UOX Fuel Fabrication

Thermal Reactors1,500 GWe






Current Burnup: 50 GWD/MTIHM:

High Burnup: 100 GWD/MTIHM:

Figure 4.1 Open Fuel Cycle: Once-Through Fuel — Projected to 2050

Depleted uranium4,430 MT/year

Separated Pu334 MT/year

Separated Uranium23,443 MT/year

MOX Fabrication Plants

PUREX Plants


Fresh MOX4,764 MTHM/year

Glass 2,886 m3/yearFP: 1,292.6 MT/yearMA: 30.1 MT/yearPu: 0.3 MT/year



1,260 Gwe from UOX240 GWe from MOX



Liquid Waste

Spent MOX4,764 MTHM/year

Figure 4.2 Closed Fuel Cycle: Plutonium Recycle (MOX option - one recycle) — Projected to 2050

MIT_ch04_29-36.qxd 7/16/2003 1:26 PM Page 30


cantly reduce the long-term radioactivity of thenuclear waste.4

In general, we expect the once-through fuelcycle to have an advantage in terms of cost andproliferation resistance (since there is no repro-cessing and separation of actinides), comparedto the closed cycle. Closed cycles have an advan-tage over the once-through cycle in terms ofresource utilization (since the recycled actinidesreduce the requirement for enriched uranium),which in the limit of very high ore prices wouldbe more economical. Some argue that closedcycles also have an advantage for long-termwaste disposal, since long-lived actinides can beseparated from the fission products and trans-muted in a reactor. Our analysis below focusedon these key comparisons.

Both once-through and closed cycles can oper-ate on U or Th fuel and can involve differentreactor types, e.g., Light Water Reactors(LWRs), Heavy Water Reactors (HWRs),Supercritical water reactors (SCWRs), HighTemperature and very High Temperature Gas

Cooled Reactors (HTGRs), Liquid Metal andGas Fast Reactors (LMFRs and GFRs), orMolten Salt Reactors (MSR) of various sizes.Today, almost all deployed reactors are of theLWR type. The introduction of new reactors orfuel cycles will require considerable develop-ment resources and some period of operatingexperience before initial deployment.

The fuel cycle characteristics of the currentworldwide deployment of nuclear power (withthe exception of three operating liquid metalfast breeder plants5) are summarized in Table4.1. At present, plants employing the once-though enriched uranium oxide (UOX) fuelhave a total capacity of about 325 GWe of elec-tricity. In addition there are plants burningreprocessed mixed Pu and U oxide fuel (MOX)in reactors with a total capacity of about 27GWe.6 Current plans call for only one recycle ofthe fuel. Table 4.1 gives the annual materialflows for the entire fleet of reactors.

The proposed mid-century deployment underthe global growth scenario of this study is

Natural uranium166,460 MT/year

Separated Uranium14,285 MT/year

MOX Fabrication Plants Pyroprocessing


Thermal Reactors815 GWe



WasteFP: 1,398 MT/yearMA+Pu: 1 MT/yearU: 551 MT/year

Fast Reactors685 GWe

Figure 4.3 Closed Fuel Cycle: Full Actinide Recycle — Projected to 2050

MIT_ch04_29-36.qxd 7/16/2003 1:26 PM Page 31


achieved either by exclusive use of the once-through cycle with current LWRs (option one)or by plutonium recycle (where all the spentUOX but none of the spent MOX isreprocessed) with current LWRs (option two).

Under both of these options, material flowsincrease significantly, as presented in Table 4.2.

The once-through fuel cycle is a technicallycredible option, assuming there is sufficienturanium ore available at reasonable cost to sup-port a deployment of this size. Note that the sin-gle-pass7 thermal reprocessing option usesalmost as much U ore as the once-through sys-tem. Furthermore, if there is adequate ore sup-ply at reasonable prices, then the single-passrecycle option will not be economically attrac-tive compared to the once through option asAppendix 4.1 discusses.

As indicated in Table 4.2, the thermal recycleoption does have an advantage in producingless material requiring permanent waste dis-posal, but this is balanced by greatertransuranic (TRU)8 waste produced duringreprocessing. Furthermore, the fission product

Table 4.1 Fuel Cycle Characteristics of Current Plantsa

U FEED 103

MT/YR HLW DISCHARGED

YR –1

Pu DISCHARGEDMT/YR

SEPARATED Pu INVENTORY MT

UOX Plants325 GWe 66.340 Spent UOX: 6471 MTIHM Discharged: 89.7 —

MOX Plants 27 GWe 3.675

Spent MOX: 179 MTIHMGlassb: 109 m3

Process Waste: 330 m3

Consumed: 12.6Discharged: 8.8 6.3c

a. Initial enrichment 4.5%, tails assay 0.3%, discharge burnup 50GWd/MTIHM, thermal efficiency 33%, capacity factor 90%. Values on a per GWe basis are given in appendix 4.b. Requires reprocessing of 944 MTIHM spent UOX per year (0.6 La Hague equivalents). Borosilicate glass contains: 48.6 MT FP, 1.1 MT Pu+MA.c. Separated Pu storage time is assumed to be 6 months. See Brogli, Krakowski, "Degree of Sustainability of Various Nuclear Fuel Cycles," Paul Scherrer Institut, August 2002.

Table 4.2 Fuel Cycle Characteristics Projected to Mid-Century

U FEED103 MT/YEAR HLW DISCHARGED YEAR –1

Pu DISCHARGEDMT/YEAR

SEPARATED Pu INVENTORY MT

Scenario 1Once-through1500 GWe

Scenario 2Thermal Recycled

UOX Plants: 780 GWeMOX Plants: 720 GWe

Scenario 1Once-through1500 GWe

Scenario 2Thermal Recycled

UOX Plants: 780 GWeMOX Plants: 720 GWe

306

257

9.45

8.18

Spent UOX: 29 864 MTIHM

Glass a: 2886 m3

Process Waste: 8785 m3

Spent MOX: 4764 MTIHM

Spent UOX: 922·103 MTIHM(13.2 YMEsc)

Spent UOX: 147·103 MTIHMSpent MOX: 124·103 MTIHM

Glassb: 75·103 m3

Process Waste: 228·103 m3

Discharged: 397

Discharged: 233

Discharged: 12.0

Discharged: 8.0

—

—

—

FLEET CUMULATIVE, FROM 352GWE IN 2002 TO 1500 GWE IN 2051

U FEED106 MT

1500 GWE FLEET PER YEAR IN 2051

HLW DISCHARGEDPu DISCHARGED

103 MT —

a. Requires reprocessing of 26 335 MTIHM spent UOX per year (14 La Hague equivalents). Borosilicate glass contains: 1292.6 MT FP, 30 MT MA, 0.3 MT Pu.b. Requires reprocessing of 651·103 MTIHM spent UOX. Borosilicate glass contains: 33.5·103 MT FP, 781 MT MA, 8.7 MT Pu.c. YME : Yucca Mountain Equivalent (70 000 MTIHM).d. MOX Plants have 2/3 of the core loaded with UOX and 1/3 loaded with MOX. Hence, 540 GWe is generated from UOX, and 240 GWe is generated from MOX.e. Separated Pu storage time is assumed to be 6 months. See Brogli, Krakowski, “Degree of Sustainability of Various Nuclear Fuel Cycles,” Paul Scherrer Institut, August 2002.

MIT_ch04_29-36.qxd 7/16/2003 1:26 PM Page 32

167e


inventory is essentially the same. Most impor-tant, the thermal recycle case has a largeamount of Pu separated each year.9 The sepa-rated plutonium inventory required for optiontwo is 167 metric tons. A nuclear weapon ofsignificant yield can comfortably be made withless than 10kg of Pu, so this amount representsthe potential for thousands of nuclearweapons. Thus, the once-through thermalrecycle scenario will not be a reasonable mid-century state, so long as U ore is available atreasonable prices. If ore prices were to becomevery high, the one-pass thermal recycle optionwould potentially be attractive, but underthose conditions, a fuel cycle that includesreactors that transmute actinides must then beconsidered (option 3). Single-pass thermalrecycle is not an attractive approach fornuclear energy for the next half century.

In option 3 we consider a fully closed fuel cycle.This fuel cycle is exactly balanced so the num-

ber of fast reactors deployed is sufficient to burnall the actinides produced in once through ther-mal reactors. Only the fast reactor fuel isreprocessed, presumably in a developed countryand a ‘secure’ energy park; the thermal reactorsoperating on a once-through cycle, can be locat-ed anywhere. This configuration has prolifera-tion advantage over the situation considered inoption two, as discussed in Chapter 8. It isimportant to note that this balanced closed fuelcycle is entirely different from breeder fast reactorfuel cycles where net plutonium produced in fastreactors is made into MOX fuel to be burned inthermal reactors. In the closed fuel cycle we con-sidered, the fast reactor burns plutonium andactinides created in the thermal reactor.

In Table 4.3, we describe three illustrativedeployments of 1500 reactors each with ratedcapacity of 1000 MWe, in order to give a moreconcrete impression of what the global growthscenario might look like. Option one is expand-

Table 4.3 Global Growth Scenario — Fuel Cycle Parameter comparison. Annual Amounts for 1500 GWe Deploymenta See Appendix 4 for fuel cycle calculations.

OPTION 1AONCE THROUGH

LOW BURN UP

OPTION 1BONCE THROUGHHIGH BURN UP

OPTION 3LWR + FAST REACTORb

LWR

Capacity, GWeEnrichment, %Burn up, GWd/MTIHMUranium ore per year, 103 MT/yr cumulative, 106 MTSpent or repr. Fuel per year, 103 MTIHM/yr cumulative, 103 MTIHMHLW, MT/yr

Pu, MT/yrWaste decay heatd

W/GWeY (100 yrs)Waste ingestion hazard m3/GWeY (1,000 yrs)

1,5004.550

3069.45

29.9922 (13.7 YME)Not applicable

397

1.1·104

6.9·1011

1,5008.2100

2868.76

14.9516 (7.4 YME)Not applicable

294

1.1·104

5.3·1011

8154.550

68525

120

Fast reactor

a. Thermal efficiency 33% for LWRs and 40% for FRs, capacity factor 90%, enrichment tails assay 0.3%. Capacity is assumed to increase linearly. Fast reactors start deployment in 15 years.b. Intended as generic fast reactor; data from ANL IFR.c. LHE means La Hague equivalent (1,700 MTHM/year)d. The decay heat and radiotoxicity are computed from and MCODE/ORIGEN run and expressed on a per GWe-y basis to establish a fair comparison between the various fuel cycles. The decay heat and radiotoxicity per unit mass can be obtained by dividing by the mass of spent fuel discharged per GWe-y. The spent fuel discharge for option 1A is 22.1 MTIHM/y, giving a decay heat at 100 years of 5.0·102 W/MTIHM and a radiotoxicity at 1000 years of 3.1·1010 m3/MTIHM, as shown in Figures 7.2 and 7.3.

1665.96

Repr.: 20.9 (12.3 LHEc)Spent : 4.1 YMEs

FP: 1398; MA+Pu: 1.0

0.7 (repr. losses)

2.8·103

2.2·107

MIT_ch04_29-36.qxd 7/16/2003 1:26 PM Page 33


ed and option two is replaced by a fully closedfuel cycle. The three options are:

Base line. 1000 MWe LWRs operating on aoncethrough fuel cycle with today’s typicalcharacteristics. (Option 1A);

Advanced once-through LWRs , perhaps withsome smaller, modular HTGR nuclear sys-tems, with higher fuel burnup characteristicsthat better meet the four objectives. (Option1B);

Fast reactors deployed in developed countrieswith a balanced closed fuel cycle.Reprocessing, fuel fabrication, and fast reac-tor burners are co-located in secure nuclearenergy “parks.” In the developing world, thedeployment is largely once-through LWRfuel cycle (Option 3).

AVAILABILITY OF URANIUM RESOURCES

How long will the uranium ore resource base besufficient to support large-scale deployment ofnuclear power without reprocessing and/orbreeding?10 Present data suggests the requiredresource base will be available at an affordablecost for a very long time. Estimates of bothknown and undiscovered uranium resources atvarious recovery costs are given in theNEA/IAEA “Red Book”11. For example, accord-ing to the latest edition of the Red Book, knownresources12 recoverable at costs < $80/kgU and< $130/kgU are approximately 3 and 4 milliontonnes of uranium, respectively. However, theamount of known resources depends on theintensity of the exploration effort, mining costs,

and the price of uranium. Thus, any predictionsof the future availability of uranium that arebased on current mining costs, prices and geo-logical knowledge are likely to be extremelyconservative.

For example, according to the AustralianUranium Information Center, a doubling of theuranium price from its current value of about$30/kgU could be expected to create about aten-fold increase in known resources recover-able at costs < $80/kgU13 i.e., from about 3 to 30million tonnes. By comparison, a fleet of 15001000 MWe reactors operating for 50 yearsrequires about 15 million tonnes of uranium(306,000 MTU/yr as indicated in Table 4.2),using conventional assumptions about burn-upand enrichment.

Moreover, there are good reasons to believe thateven as demand increases the price of uraniumwill remain relatively low: the history of allextractive metal industries, e.g., copper, indi-cates that increasing demand stimulates thedevelopment of new mining technology thatgreatly decreases the cost of recovering addi-tional ore. Finally, since the cost of uraniumrepresents only a small fraction of the busbarcost of nuclear electricity, even large increases inthe former — as may be required to recover thevery large quantities of uranium contained atlow concentrations in both terrestrial depositsand seawater — may not substantially increasethe latter.14 In sum, we conclude that resourceutilization is not a pressing reason for proceed-ing to reprocessing and breeding for many yearsto come.

MIT_ch04_29-36.qxd 7/16/2003 1:26 PM Page 34


NOTES

1. See, for example, OECD Nuclear Energy Agency, Trends inthe Nuclear Fuel Cycle ISBN 92-64-19664-1 (2001) andNuclear Science Committee “Summary of the workshopon advanced reactors with innovative fuel,” October1998, NEA/NSC/DOC(99)2.

2. Several nations have explored breeder reactors, notablythe U.S., France, Russia, Japan, and India.

3. Minor actinides are Americium (Am),Neptunium (Np),and Curium (Cm).

4. There are still other options, such as using an acceleratorto produce neutrons in a sub-critical assembly.

5. The three surviving developmental breeder reactors arePhenix in France, Monju in Japan, and BN600 in Russia.

6. The MOX fueled plants are currently operating with onlyabout a third of their core loaded as MOX fuel; the bal-ance is UOX fuel. Hence only about 9 GWe are beinggenerated in these reactors from the MOX fuel

7. Single pass recycle means that a discharged fuel batch isreprocessed once only.

8. TRU here refers to the U.S. definition: low-level wastecontaminated with transuranic elements.

9. Due to process holding time, the actual amount of sepa-rated Pu inventory could be several or more years’ worthof separations.

10. For additional details, see Appendix 5-E and MarvinMiller, Uranium resources and the future of nuclearpower, Lecture notes, MIT, Spring 2001; for copies con-tact [email protected].

11. Uranium resources, production, and demand (“The RedBook”), OECD Nuclear Energy Agency and InternationalAtomic Energy Agency, 2001.

12. Such resources are also known as measured resourcesand reserves.

13. Uranium Information Center,“Nuclear Electricity”, 6th

edition, Chapter 3 (2000). Available on the web athttp:www.uic.com.au/ne3.htm.

14. For example, recent research in Japan indicates that uranium in seawater — present in concentration of 3.3 ppb — might be recovered at costs in the range of $300–$500/kg.

MIT_ch04_29-36.qxd 7/16/2003 1:26 PM Page 35

MIT_ch04_29-36.qxd 7/16/2003 1:26 PM Page 36

37C h a p t e r 5 — N u c l e a r P o w e r E c o n o m i c s

Investments in commercial nuclear generatingfacilities will only be forthcoming if investorsexpect the cost of producing electricity usingnuclear power will be lower than the risk-adjusted costs associated with alternative elec-tric generation technologies. Since nuclearpower plants have relatively high capital costsand very low marginal operating costs, nuclearenergy will compete with alternative electricitygeneration sources for “baseload” (high loadfactor) operation. We recognize that over thenext 50 years some significant but uncertainfraction of incremental electricity supplies willcome from renewable energy sources (e.g.wind) either because these sources are less cost-ly than alternatives or because governmentpolicies (e.g. production tax credits, high man-dated purchase prices, and renewable energyportfolio standards) or consumer choice favorrenewable energy investments. Despite theefforts to promote renewable energy options,however, it is likely that a large fraction of theincremental and replacement investments inelectric generating capacity needed to balancesupply and demand over the next 50 years will,in the absence of a nuclear generation option,rely on fossil-fuels — primarily natural gas orcoal. This is particularly likely in developingcountries experiencing rapid growth in incomeand electricity consumption. Accordingly, wefocus on the costs of nuclear power comparedto these fossil fuel generating alternatives inbase-load applications.

Any analysis of the costs of nuclear power musttake into account a number of important con-siderations. First, all of the nuclear power plantsoperating today were developed by state-ownedor regulated investor-owned vertically-integrat-

ed utility monopolies.1 Many developed coun-tries and an increasing number of developingcountries are in the process of moving awayfrom an electric industry structure built uponvertically integrated regulated monopolies toan industry structure that relies primarily oncompetitive generation power plant investors.We assume that in the future nuclear power willhave to compete with alternative generatingtechnologies in competitive wholesale markets— as merchant plants.2 These changes in thestructure of the electric power sector haveimportant implications for investment in gen-erating capacity. Under traditional industry andregulatory arrangements, many of the risksassociated with construction costs, operatingperformance, fuel price changes, and other fac-tors were borne by consumers rather than sup-

pliers.3 The insulation of investors from manyof these risks necessarily had significant effectson the cost of capital they used to evaluatealternative generation options and on whetherand how they took extreme contingencies intoaccount. Specifically, the process reduced thecost of capital and led investors to give lessweight to regulatory (e.g. construction andoperating licenses) and construction costuncertainty, operating performance uncertain-ties and uncertainties associated with future oil,gas and coal prices than if they had to bear thesecost and performance risks.

In a competitive generation market it isinvestors rather than consumers who must bearthe risk of uncertainties associated with obtain-ing construction and operating permits, con-struction costs and operating performance.While some of the risks associated with uncer-tainties about the future market value of elec-

Chapter 5 — Nuclear Power Economics

MIT_ch05_37-46.qxd 7/16/2003 1:29 PM Page 37


plants to about $13/MWe-hr by 2001)8, ratherthan the $10/MWe-hr often assumed in manypaper engineering cost studies.

Third, even if an investment in nuclear powerlooked attractive on a spreadsheet, investorsmust confront the regulatory and political chal-lenges associated with obtaining a license tobuild and operate a plant on a specific site. Inthe past, disputes about licensing, local opposi-tion, cooling water source and dischargerequirements, etc., have delayed constructionand completion of nuclear plants. Manyplanned plants, some of which had incurredconsiderable development costs, were can-celled. Delays and “dry-hole” costs are especial-ly burdensome for investors in a competitiveelectricity market.

With these considerations in mind, we nowproceed to examine the relative costs of newnuclear power plants, pulverized coal plants,and combined-cycle gas turbine (CCGT) plantsin base-load operations in the United States.9

The analysis is not designed to produce preciseestimates, but rather a “reasonable” range ofestimates under a number of different assump-tions reflecting uncertainties about future con-struction and operating costs. Similar analysisfor Europe and especially Japan and Koreawould be somewhat more favorable to nuclear,since gas and coal costs are typically higher thanin the United States.

We start with a “base case” that examines thelevelized real life-cycle costs of nuclear, coal,and CCGT generating technology usingassumptions that we believe commercialinvestors would be expected to use today toevaluate the costs of the alternative generationoptions. The levelized cost is the constant realwholesale price of electricity that meets a pri-vate investor’s financing cost, debt repayment,income tax, and associated cash flow con-straints.

The base case assumes that non-fuel O&M costscan be reduced by about 25% compared to therecent operating cost experience of the average

tricity can be shifted to electricity marketersand consumers through forward contracts,some market risk and all construction cost,operating cost and performance risks will con-tinue to be held by power plant investors.4

Thus, the shift to a competitive electricity mar-ket regime necessarily leads investors to favorless capital-intensive and shorter constructionlead-time investments, other things equal.5 Itmay also lead investors to favor investmentsthat have a natural “hedge” against market pricevolatility, other things equal.6

Second, the construction costs of nuclear plantscompleted during the 1980s and early 1990s inthe United States and in most of Europe werevery high — and much higher than predictedtoday by the few utilities now building nuclearplants and by the nuclear industry generally.The reasons for the poor historical constructioncost experience are not well understood andhave not been studied carefully. The realizedhistorical construction costs reflected a combi-nation of regulatory delays, redesign require-ments, construction management and qualitycontrol problems. Moreover, construction onfew new nuclear power plants has been startedand completed anywhere in the world in thelast decade. The information available aboutthe true costs of building nuclear plants inrecent years is also limited. Accordingly, thefuture construction costs of building a largefleet of nuclear power plants is necessarilyuncertain, though the specter of high construc-tion costs has been a major factor leading tovery little credible commercial interest ininvestments in new nuclear plants. Finally,while average U.S. nuclear plant availability hasincreased steadily during the 1990s to a high of90% in 2001, many nuclear plants struggledwith low availabilities for many years and thelife-cycle availability of the fleet of nuclearplants (especially taking account of plants thatwere closed early) is much less than 90%.7 Inaddition, the average operation and mainte-nance costs of U.S. nuclear plants (including

(though average O&M costs had fallen to about$18/MWe-hr and the lowest cost quartile of

MIT_ch05_37-46.qxd 7/16/2003 1:29 PM Page 38

fuel) were over $20/MWh during the 1990s


nuclear plant operating in the U.S. in the lastfew years. This puts the total O&M costs(including fuel) at about 15 mills/kWe-hr. Weinclude this reduction in O&M costs in the basecase because we expect that operators of newnuclear plants in a competitive wholesale elec-tricity market environment will have to demon-strate better than average performance toinvestors. The 15 mill O&M cost value is consis-tent with the performance of existing plantsthat fall in the second lowest cost quartile ofoperating nuclear plants.10 (The assumptionsunderlying the base case are listed in Table 5.3and illustrative cash flows produced by ourfinancial model are provided in Appendix 5.)

We then examine how the real levelized cost ofnuclear generated electricity changes as weallow for additional cost improvements. First,we assume that construction costs can bereduced by 25% from the base case levels tomore closely match optimistic but plausibleforecasts. Second, we examine how life-cyclecosts are further reduced by a one-year reduc-tion in construction time. Third, we examinethe effects of reducing financing costs to a levelcomparable to what we assume for gas and coalgenerating units as a consequence of, for exam-ple, reducing regulatory risks and commercialrisks associated with uncertainties about con-struction and operating costs that presentlyburden nuclear compared to fossil-fueled alter-natives. This reduction in financial risk mightresult from an effective commercial demonstra-tion program of the type that we discuss furtherin Part II. Finally, we examine how the relativecosts of coal and CCGT generation are affectedby placing a “price” on carbon emissions,through carbon taxes, the introduction of a car-bon emissions cap and trade program, or equiv-alent mechanism to price carbon emissions tointernalize their social costs into investmentdecisions in a way that treats all supply optionson an equivalent basis. We consider carbonprices in a range that brackets current estimatesof the costs of carbon sequestration (capture,transport and storage). The latter analysis pro-vides a framework for assessing the option valueof nuclear power if and when the United States

adopts a program to stabilize and then reducecarbon emissions.

The levelized cost of electric generating plantshas typically been calculated under the assump-tion that their regulated utility owners recovertheir costs using traditional regulated utilitycost of service cost recovery rules. Investmentswere recovered over a 40 year period and debtand equity were repaid in equal proportionsover this lengthy period at the utility’s cost ofcapital, which reflected the risk reducing effectsof regulation. Moreover, the calculations typi-cally provided levelized nominal cost valuesrather than levelized real cost values, obscuringthe effects of inflation and making capitalintensive technologies look more costly relativeto alternatives than they really were.

We do not believe that these traditional lev-elized cost models based on regulated utilitycost recovery principles provide a good descrip-tion of how merchant plants will be financed inthe future by private investors. Accordingly, wehave developed and utilized an alternativemodel that provides flexibility to specify morerealistic debt repayment obligations and associ-ated cash flow constraints, as well as the costs ofdebt and equity and income tax obligations thata private firm would assign to individual proj-ects with specific risk attributes, while account-ing for corporate income taxes, tax depreciationand the tax shield on interest payments. Werefer to this as the Merchant Cash Flow model.We have relied primarily on simulation resultsusing this model under assumptions of both a25-year and 40-year capital recovery period and85% and 75% lifetime capacity factors.

BASE CASE

The base case reflects reasonable estimates ofthe current perceived costs of building andoperating the three generating alternatives in2002 U.S. dollars. The overnight capital cost for

cussed in Appendix 5, this value is consistentwith estimates made by the U.S. Energy

MIT_ch05_37-46.qxd 7/16/2003 1:29 PM Page 39

nuclear in the base case is $2000/kWe. As dis-


Information Administration (EIA), estimatesreported by other countries to the OECD, andrecent nuclear plant construction experienceabroad. We have not relied on construction costdata for U.S. plants completed in the late 1980sand early 1990s; if we had, the average overnightconstruction cost in 2002 U.S. dollars wouldhave been much higher. We are aware that somevendors and some potential investors in newnuclear plants believe that they can achievemuch lower construction costs. We considersignificant construction cost reductions in ourdiscussion of improvements in nuclear costs.11

As previously discussed, our base case assumesthat O&M costs are 15 mills/kWe-hr, which islower than the recent experience for the averagenuclear plant and is consistent with the recentperformance of plants in the second lowest costquartile of operating nuclear plants in the U.S.The O&M costs of plants in the lowest costquartile (best performers) are about 13mills/kWe-hr. We consider this to represent thepotential for further cost improvements for afleet of new nuclear plants but we do not believethat investors will assume that all plants willachieve the O&M cost levels of the best per-formers.

The construction costs assumed for CCGT andcoal plants are in line with experience and EIAestimates. The construction cost of the coalplant is assumed to reflect NOx and SO2 con-trols as required to meet current New SourcePerformance Standards. There are four casespresented for the CCGT plants: (1) a low gasprice case that starts with gas prices at$3.50/MMBtu which rise at a real rate of 0.5%over 40 years (real levelized cost of$3.77/MMbtu over 40 years); (2) a moderategas price case with gas prices starting at$3.50/MMBtu as well, but rising at a real rate of1.5% per year over 40 years (real levelized costof $4.42 over 40 years); (3) high gas price casethat starts at $4.50/MMbtu and rises at a realrate of 2.5% per year (real levelized cost of$6.72/Mmbtu over 40 years). (4) The fourthCCGT case reflects high gas prices and anadvanced CCGT design with a (roughly) 10%

improvement in its heat rate. The base caseresults for 25 and 40-year economic lives and85% capacity factor are reported in Table 5.1and the equivalent results for a 75% lifetimecapacity factor are reported in Table 5.2. Theassumptions for the cases are given in Table 5.3.The discussion that follows is based on the 85%capacity factor simulations since the basicresults don’t change very much when weassume the lower capacity factor.

The base case results suggest that nuclear poweris much more costly than the coal and gas alter-natives even in the high gas price cases. In thelow gas price case, CCGT is cheaper than coal.In the moderate gas price case, total life-cyclecoal and gas costs are quite close together,though we should recognize that there areregions of the country with below average coalcosts where coal would be less costly than gasand vice versa. Under the high gas priceassumption, coal beats gas by a significantamount. (We have not tried to account for therelative difficulties of siting coal and gas plants.)We discuss potential future carbon emissionsregulations separately below.

This suggests that high natural gas prices willeventually lead investors to switch to coal ratherthan to nuclear under the base case assump-tions as nuclear appears to be so much morecostly than coal and U.S. coal supplies are veryelastic in the long run so that significantincreases in coal demand will not lead to signif-icant increases in long term coal prices. Incountries with less favorable access to coal, thegap would be smaller, but 2.5 cents/kWe-hr istoo large a gap for nuclear to beat coal in manyareas of the world under the base case assump-tions (absent additional restrictions on emis-sions of carbon dioxide from coal plants whichwe examine separately below).

The bottom line is that with current expecta-tions about nuclear power plant constructioncosts, operating cost and regulatory uncertain-ties, it is extremely unlikely that nuclear powerwill be the technology of choice for merchantplant investors in regions where suppliers have

MIT_ch05_37-46.qxd 7/16/2003 1:29 PM Page 40


access to natural gas or coal resources. It is justtoo expensive. In countries that rely on stateowned enterprises that are willing and able toshift cost risks to consumers to reduce the costof capital, or to subsidize financing costs direct-ly, and which face high gas and coal costs, it ispossible that nuclear power could be perceivedto be an economical choice.12

IMPROVEMENTS IN NUCLEAR COSTS

We next examine how the cost of electricitygenerated by nuclear power plants wouldchange, if effective actions can be taken toreduce nuclear electric generation costs in sev-eral different ways. First, we assume that con-struction costs can be reduced by 25%. Thisbrings the construction costs of a nuclear plantto a level more in line with what the nuclearindustry believes is feasible in the medium termunder the right conditions.13 While this reducesthe levelized cost of nuclear electricity consider-ably, it is still not competitive with gas or coalfor any of the base cases. Reducing constructiontime from 5 years to 4 years reduces the lev-elized cost further, but not to a level that wouldmake it competitive with fossil fuels. However, ifregulatory, construction and operating costuncertainties could be resolved, and the nuclearplant could be financed under the same termsand conditions (cost of capital) as a coal or gasplant, then the costs of nuclear power becomevery competitive with the costs of CCGTs in ahigh gas price world and only slightly morecostly than pulverized coal plants, assumingthat comparable improvements in the costs ofbuilding coal plants are not also achieved. Ifnuclear plant operators could reduce O&Mcosts by another 2 mills to 13 mills/kWe-hr,consistent with the best performers in theindustry, nuclear’s total cost would match thecost of coal and the cost of CCGT in the mod-erate and high gas price cases. However, nucleardoes not have a meaningful economic advan-tage over coal.

These results suggest that with significantimprovements in the costs of building, operat-

ing, and financing nuclear power plants, andcontinued excellent operating performance(85% capacity factor), nuclear power could bequite competitive with natural gas if gas pricesturn out to be higher than what most analystsnow appear to believe and would be only slight-ly more costly than coal within the range ofassumptions identified.14

The cost improvements we project are plausiblebut unproven. It should be emphasized, that thecost improvements required to make nuclearpower competitive with coal are significant:25% reduction in construction costs; greaterthan a 25% reduction in non-fuel O&M costscompared to recent historical experience(reflected in the base case), reducing the con-struction time from 5 years (already optimistic)to 4 years, and achieving an investment envi-ronment in which nuclear power plants can befinanced under the same terms and conditionsas can coal plants. Moreover, under what weconsider to be optimistic, but plausible assump-tions, nuclear is never less costly than coal.

CARBON “TAXES”

From a societal cost perspective, all externalsocial costs of electricity generation should bereflected in the price. Here we consider the costof CO2 emissions and not other externalities;for example we ignore the costs of other air pol-lutants from fossil fuel combustion and nuclearproliferation and waste issues (except forincluding the costs of new coal plants to meetnew source performance standards). Nuclearlooks more attractive when the cost of CO2

emissions is taken into account. Unlike gas andcoal-fired plants, nuclear plants produce no car-bon dioxide during operation and do not con-tribute to global climate change. Accordingly, itis natural to explore what the comparativesocial cost of nuclear power would be, if carbonemissions were “priced” to reflect the marginalcost of achieving global carbon emissions stabi-lization and reduction targets.15 Future UnitedStates policies regarding carbon emissions areuncertain at the present time.

MIT_ch05_37-46.qxd 7/16/2003 1:29 PM Page 41


By examining the relative economics of nuclearpower under different assumptions aboutfuture social valuations for reducing carbonemissions, we can get a feeling for the optionvalue of nuclear generation in a world with car-bon emissions restrictions of various severities.

To examine this question we have recalculatedthe costs of the fossil-fueled generation alterna-tives to reflect a carbon tax of $50/tC, $100/tC,and $200/tC. The lower value is consistent withan EPA estimate of the cost of reducing U.S.CO2 emissions by about 1 billion metric tonsper year.16 The $100/tC and $200/tC valuesbracket the range of values that appear in theliterature regarding the costs of carbon seques-tration, recognizing that there is enormousuncertainty about the costs of deploying CO2

capture, transport, and storage on a large scale.These hypothetical taxes should be thought ofas a range of “backstop” marginal costs forreducing carbon emissions to meet aggressiveglobal emissions goals. These results are report-ed in Table 5.1 and 5.2, as well.

With carbon taxes in the $50/tC range, nuclearis not economical under the base case assump-tions. If nuclear costs can be reduced to reflectall of the cost-reduction specifications dis-cussed earlier, nuclear would be less costly thancoal and less costly than gas in the high gas pricecases. It is roughly competitive with gas in thelow and moderate price gas cases. With carbontaxes in the $100/tC to $200/tC range, nuclearpower would be an economical base loadoption compared to coal under the base caseassumptions, but would still be more costlythan gas except in the high gas price case.However, nuclear would be significantly lesscostly than all of the alternatives with carbonprices at this level, if all of the cost reductionspecifications discussed earlier could beachieved.

The last conclusion ignores one important con-sideration. With carbon taxes at these high lev-els, it could become economical to deploy agenerating technology involving the gasificationof coal, its combustion in a CCGT (IGCC), and

Table 5.1 Costs of Electric Generation AlternativesReal Levelized Cents/kWe-hr (85% capacity factor)

Base Case 25-YEAR 40-YEAR

NuclearCoalGas (low)Gas (moderate)Gas (high)Gas (high) Advanced

7.04.43.84.15.34.9

6.74.23.84.15.65.1

Reduce Nuclear Costs Cases

Reduce construction costs (25%). Reduce construction time by 12 months Reduce cost of capital to be equivalent to coal and gas

5.85.6

4.7

5.55.3

4.4

Carbon Tax Cases (25/40 year)

$50/tC $100/tC $200/tC

CoalGas (low)Gas (moderate)Gas (high)Gas (high) advanced

5.6/5.44.3/4.34.6/4.75.8/6.15.3/5.6

6.8/6.64.9/4.85.1/5.26.4/6.75.8/6.0

9.2/9.05.9/5.96.2/6.27.4/7.76.7/7.0

Table 5.2 Costs of Electric Generation AlternativesReal Levelized Cents/kWe-hr (75% capacity factor)

Base Case 25-YEAR 40-YEAR

NuclearCoalGas (low)Gas (moderate)Gas (high)Gas (high) advanced

7.94.84.04.25.55.0

7.54.63.94.35.75.2

Reduce Nuclear Costs Cases

Reduce construction costs (25%) Reduce construction time by 12 months Reduce cost of capital to be equivalent to coal and gas

6.56.2

5.2

6.26.0

4.9

Carbon Tax Cases (25/40 year)

$50/tC $100/tC $200/tC

CoalGas (low)Gas (moderate)Gas (high)Gas (high) advanced

6.0/5.84.5/4.44.7/4.86.0/6.35.5/5.7

7.2/7.05.0/5.05.3/5.36.5/6.85.9/6.2

9.6/9.46.0/6.06.3/6.47.5/7.86.8/7.1

MIT_ch05_37-46.qxd 7/16/2003 1:29 PM Page 42


the sequestration of carbon dioxide producedin the process. The potential cost savings fromthis technology compared to conventional pul-verized coal plants arises from (a) the use of rel-atively inexpensive coal to produce syngas(mostly CO and H2) (b) the higher thermal effi-ciency of CCGT, and more economical captureof CO2. Depending on the economics of thistechnology, coal could play a larger competitiverole in a world with high carbon taxes thanmight be suggested by Tables 5.1 and 5.2. Weobserve as well, that from an environmentalperspective, the world looks very different ifthere are abundant supplies of cheap naturalgas, than if natural gas supplies are scarcer andsignificantly more expensive than many recentprojections imply.

INTERNATIONAL PERSPECTIVE ON COST OFELECTRICITY

The methodology followed above is pertinentto an electricity generation market that isunregulated, a situation that the United States ismoving toward, as are several other countries.An additional advantage to describing deregu-lated market situations is that the methodologyproperly focuses on the true economic cost ofelectricity generating alternatives. There arehowever many nations that do not enjoy anunregulated generating market and are unlikelyto adopt deregulation for some time to come. Inmany of these countries electricity generation isrun directly or indirectly by the governmentand significant subsidies are provided to gener-ating facilities. The electricity “cost” in thesecountries is not transparent and leads to a dif-ferent political attitude toward investment deci-sions because consumers enjoy subsidizedprices. The result is a misallocation of resourcesand over the long-run one can expect that polit-ical and economic forces will call for change.These non-market situations are encounteredin Europe, e.g. Electricite de France, althoughthere is a strong move to deregulation in the EUand in developing countries that frequentlyhave state run power companies. Importantly,the costs of advanced fuel cycle technologies

Table 5.3 Base Case Assumptions

NuclearOvernight cost: $2000/kWeO&M cost: 1.5 cents/kWh (includes fuel)O&M real escalation rate: 1.0%/yearConstruction period: 5 yearsCapacity factor: 85%/75%Financing: Equity: 15% nominal net of income taxes Debt: 8% nominal Inflation: 3% Income Tax rate (applied after expenses, interest and tax depreciation): 38% Equity: 50% Debt: 50%Project economic life: 40 years/25 years

CoalOvernight cost: $1300/kWeFuel Cost: $1.20/MMbtuReal fuel cost escalation: 0.5% per yearHeat rate (bus bar): 9300 BTU/klWhConstruction period: 4 yearsCapacity factor: 85%/75%Financing: Equity: 12% nominal net of income taxes Debt: 8% nominal Inflation: 3% Income Tax rate (applied after expenses, interest and tax depreciation): 38% Equity: 40% Debt: 60%Project economic life: 40 years/25 years

Gas CCGTOvernight cost: $500/kWeInitial fuel cost: Low: $3.50/MMbtu ($3.77/MMbtu real levelized over 40 years) Moderate: $3.50/MMbtu ($4.42/MMbtu real levelized over 40 years) High: $4.50/MMbtu ($6.72/MMbtu real levelized over 40 years)Real fuel cost escalation: Low: 0.5% per year Moderate: 1.5% per year High: 2.5% per yearHeat rate: 7200 BTU/kWh Advanced: 6400 BTU/kWhConstruction period: 2 yearsCapacity factor: 85%/75%Financing: Equity: 12% nominal net of income taxes Debt: 8% nominal Inflation: 3% Income tax rate (applied after expenses, interest and tax depreciation): 38% Equity: 40% Debt: 60%Project economic life: 40 years/25 years

MIT_ch05_37-46.qxd 7/16/2003 1:29 PM Page 43


such as PUREX reprocessing and MOX fabrica-tion are heavily subsidized reflecting politicalrather than economic decision making.

COST OF ADVANCED FUEL CYCLES.

We have not undertaken as complete analysisfor the costs of advanced fuel cycles as we havefor the open fuel cycle. We have however exam-ined in some detail the cost of the closed fuelcycle with single pass PUREX/MOX relative tothe open cycle. This analysis is reported in theAppendix 5.D.

The fuel cycle cost model presented inAppendix 5.D shows that the closed cyclePUREX/MOX option fuel costs are roughly 4times greater than for the open cycle, using esti-mated costs under U.S. conditions. The closedcycle can be shown to be competitive with theonce-through option only if the price of urani-um is high and if optimistic assumptions aremade regarding the cost of reprocessing, MOXfabrication, and high level waste disposal. Asexplained in Appendix 5.D, the effect of theincreased MOX fuel cycle cost on the cost ofelectricity depends upon the percentage ofMOX fuel in the entire fleet if fuel costs areblended.

The case is often advanced that disposing ofreprocessed high level waste will be less expen-sive than disposing of spent fuel directly. Butthere can be little confidence today in any esti-mate of such cost savings, especially if disposalof non-high-level waste contaminated with sig-nificant quantities of long-lived transuranicradionuclides (TRU waste) associated withrecycle facilities and operations is taken intoaccount. Furthermore, our cost model showsthat even if the cost of disposing of reprocessedhigh-level waste were zero, the basic conclusionthat reprocessing is uneconomic would notchange.

It should be noted that the cost incrementassociated with reprocessing and thermal recy-cle is small relative to the total cost of nuclearelectricity generation. In addition, the uncer-tainty in any estimate of fuel cycle costs isextremely large.

NOTES

1. Though in the United States and the United Kingdomsome nuclear plants were subsequently sold or trans-ferred to merchant generating companies.

2. Merchant plants sell their output under short, mediumand longer term supply contracts negotiated competi-tively with distribution companies, wholesale and retailmarketers. The power plant developers take on permit-ting, development, construction cost and operating per-formance risks but may transfer some or all risks associ-ated with market price volatility to buyers (for a price)through the terms of their contracts.

3. It is often assumed that regulated monopolies were sub-ject to “cost-plus” regulation which insulated utilitiesfrom all of these risks. This is an extreme and inaccuratecharacterization of the regulatory process, at least in theUnited States. (P.L. Joskow and R. Schmalensee,“IncentiveRegulation for Electric Utilities,” Yale Journal onRegulation, 1986; P.L. Joskow,“Deregulation andRegulatory Reform in the U.S. Electric Power Sector,” inDeregulation of Network Industries: The Next Steps (S.Peltzman and Clifford Winston, eds.), Brookings Press,2000 ). Several U.S. utilities were faced with significantcost disallowances associated with nuclear power plantsthey completed or abandoned, a result inconsistent withpure cost-plus regulation. Nevertheless, it is clear that alarge fraction of these cost and market risks were shiftedto consumers from investors when the industry wasgoverned by regulated monopolies.

4. The current state of electricity restructuring and compe-tition in the United States and Europe has made it diffi-cult for suppliers to obtain forward contracts for thepower they produce. We believe that this chaotic situa-tion is unsustainable and that a mature competitivepower market will make it possible for power suppliersto enter into forward contracts with intermediaries.However, these contracts will not generally be like the30-year contracts that emerged under regulation whichobligated wholesale purchasers (e.g. municipal utilities)to pay for all of the costs of a power plant in return forany power it happened to produce. In a competitivemarket the contracts will be for specified delivery obli-gations at a specified price (or price formula), will tendto be much shorter (e.g. 5-year contract portfolios), andwill place cost and operating performance risk on thegenerator not on the customer.

MIT_ch05_37-46.qxd 7/16/2003 1:29 PM Page 44


5. Oversimplifying, these effects can be thought of as anincrease in the cost of capital faced by investors.

6. For example, in areas of the United States where thewholesale market tends to clear with conventional gasor oil-fired power plants on the margin, spot marketclearing prices will move up and down with the price ofnatural gas and oil. A combined cycle gas turbine(CCGT) that also burns natural gas, but with a heat rate35% lower on average than those of the marginal gasplants that clear the market (e.g. 11,000 BTU/kWh), willalways run underneath the market clearing price of elec-tricity. Whatever the price of gas, the CCGT is always inthe money and will be economical to run under thesecircumstances. If gas prices go up, the CCGT will bemore profitable, and if they go down it will be less prof-itable, but the volatility in profits with respect tochanges in gas prices will be lower than that for coal ornuclear plants.

7. In 2000, the capacity factors for the nuclear plants inFrance were 76%, for those in Japan 79%, and for thosein South Korea, 91%. Ideally, we would look at availabili-ty data, but except for France where nuclear accountsfor such a large share of electricity supply that someplants must by cycled up and down, nuclear units aregenerally run full out when they are available (Source:Calculated from data on EIA web site.)

8. These numbers underestimate the true O&M costs ofnuclear plants because they exclude administrative andgeneral operating costs that are typically captured else-where in utility income statements. These overheadcosts probably add another 20% to nuclear O&M costs.We do not consider these additional costs here becausethey are also excluded from the O&M costs for compet-ing technologies. In a competitive power market, how-ever, generating plants must earn enough revenues tocover these overhead costs as well as their direct capitaland O&M costs.

9. That is, we are not considering competition betweennew nuclear plants and existing coal and gas plants(whose construction costs are now sunk costs). We rec-ognize there may be economical opportunities toincrease the capacity of some existing nuclear plantsand to extend their commercial lives. We do not consid-er these opportunities here.

10. The reduced non-fuel O&M costs assumed are about 10mills/kWh in the base case and compare favorably to 9mills/kWh assumed by TVA (90% capacity factor) in itsrecent evaluation of the restart of Browns Ferry Unit #1.

11. Of course, in a competitive wholesale electricity marketinvestors are free to act on such expectations by makingfinancial commitments to build new nuclear plants.

been built in the U.S. in the last five years, most of itowned by merchant investors and most of it fueled bynatural gas and none of it nuclear. See Paul L. Joskow,“The Difficult Transition to Competitive ElectricityMarkets in the U.S.,” May 2003

12. We have seen some analyses that assume that nuclearplants will be financed with 100% government-backeddebt, pay no income or property taxes, and have verylong repayment schedules. One can make the costs ofnuclear power look lower this way, but it simply hidesthe true costs and risks of the projects which have effec-tively been transferred to consumers and taxpayers.

13. This brings the nuclear plant cost down to $1500/kW.This is roughly the cost used in the analysis of the costsof a new nuclear power plant in Finland at currentexchange rates. (However, the Finnish analysis assumesthat the plant can be financed with 100% debt at a 5%real interest rate and would pay no income taxes). Note,however, that TVA estimates that the costs of refurbish-ing a mothballed unit at Browns Ferry will cost about$1300/kWe, and that recent Japanese experience is clos-er to the $2000/kWe base case assumption. TVA’s analy-sis of the costs of refurbishing the Browns Ferry unitassume that the project can be financed with 100%debt at an interest rate 80 basis points above 10-yeartreasury notes and would pay no taxes.

14. Obviously, there is some set of assumptions that willmake nuclear cheaper than coal. However, they basicallyrequire driving the construction costs and constructiontime profile to be roughly equivalent to those of a coalunit. We also have not assumed any improvements inconstruction costs or heat rates for coal units associatedwith advanced coal plant designs.

15. We have modeled the carbon “price” as a carbon dioxideemissions tax. However, the intention is to simulate anypolicies that give nuclear power “credit” relative to fossilfuel alternatives for producing no CO2.

16. “Summary and Analysis of McCain-Leiberman ‘ClimateStewardship Act of 2003’,”William Pizer and RaymondKopp, Resources for the Future, January 28, 2003.

MIT_ch05_37-46.qxd 7/16/2003 1:29 PM Page 45

About 150,000 MWe of new generating capacity has

MIT_ch05_37-46.qxd 7/16/2003 1:29 PM Page 46

47C h a p t e r 6 — S a f e t y

Safe operations of the entire nuclear fuel cycleare a paramount concern. In this chapter weaddress reactor safety, the continuing availabil-ity of trained personnel for nuclear operations,the threat of terrorist attack, and nuclear fuelcycle safety, including nuclear fuel reprocessingplants.

There are about 100 nuclear power plants in theU. S., and over 400 in the world, mostly lightwater reactors (LWRs). With the benefit ofexperience and improved plant designs goinginto service, performance has improved overtime to unit capacity factors1 of 90% and high-er in the U.S.2 The means of improvementinclude independent peer review and the feed-back of operating experience at reactor fleetsworldwide, so that all operators become awareof mishaps that occur, and the commitment ofplant owners and managements to the develop-ment of safety culture within the organizationsthat operate nuclear power plants. Theseactions and initiatives in training and qualifica-tion of reactor operators that have been imple-mented by organizations of plant owners3 aremajor factors in the performance improve-ments. Experience also includes three seriousreactor accidents4 and several fuel cycle facilityaccidents.5

A number of events have occurred at reactorsthat were headed for an accident but stoppedshort. Such an event6 came to light during aninspection of the Davis-Besse reactor vesselhead in March, 2002, during reactor shutdown.The inspection disclosed a large cavity in thevessel head next to one of the reactor controlrod drive mechanisms, caused by boric acidleakage and corrosion. The cavity seriously

jeopardized reactor vessel integrity. Fortunately,the fault was discovered before restart of thereactor. This event discloses a failure on the partof the plant owners to respond to earlier indica-tions of an issue and to look for problems in anearly stage at their plant. It is still an open ques-tion whether the average performers in theindustry have yet incorporated an effective safe-ty culture into their conduct of business. TheU.S. Nuclear Regulatory Commission sharesresponsibility in the matter, as it accepted delayof scheduled surveillance and inspection ofvital primary system components. A majornuclear power initiative will not gain publicconfidence, if such failures occur.

With regard to the mandate of the NuclearRegulatory Commission for safety of nuclearplants in the U. S., the Davis-Besse incident alsoraises questions about whether nuclear reactorsafety goals are compatible with the transitionto competitive electricity markets. On the onehand some observers suggest that unregulatedgenerators will be more concerned with maxi-mizing plant output and less willing to closeplants for safety inspections and correctiveactions where necessary. On the other hand,owners groups have long stated that nuclearplant operation conducted to ensure a highlevel of safety is also economically beneficial.Further, nuclear plant accident costs are notfinancially attractive for plant owners. Whilethere may be some accident costs that are notfully internalized into decisions made by indi-vidual nuclear plant owners, the owner of aplant that has a serious accident would face verysignificant adverse financial consequences, aswas the case of General Public Utilities after theaccident at Three Mile Island Unit 2. We believe

Chapter 6 — Safety

MIT_ch06_47-52.qxd 7/16/2003 1:31 PM Page 47


damage frequency of U.S. reactors is therefore 1in 2679 reactor-years on average.

Probabilistic Risk Assessment (PRA) identifiespossible failures that can occur in the reactor,e.g., pipe breaks or loss-of-reactor coolant flow,then traces the sequences of events that follow,and finally determines the likelihood of theirleading to core damage. PRA includes bothinternal events and external events, i.e., naturaldisasters. Expert opinion using PRA considersthe best estimate of core damage frequency tobe about 1 in 10,000 reactor-years for nuclearplants in the United States. Although safetytechnology has improved greatly with experi-ence, remaining uncertainties in PRA methodsand data bases make it prudent to keep actualhistorical risk experience in mind when makingjudgments about safety.

With regard to implementation of the globalgrowth scenario during the period 2005-2055,both the historical and the PRA data show anunacceptable accident frequency. The expectednumber of core damage accidents during thescenario with current technology8 would be 4.We believe that the number of accidents expect-ed during this period should be 1 or less, whichwould be comparable with the safety of the cur-rent world LWR fleet. A larger number posespotential significant public health risks and, asalready noted, would destroy public confidence.We believe a ten-fold reduction in the likeli-hood of a serious reactor accident,9 i.e., a coredamage frequency of 1 in 100,000 reactor- yearsis a desirable goal and is also possible, based onclaims of advanced LWR designers, that webelieve plausible. In fact, advanced LWRdesigners claim that their plant designs alreadymeet this goal, with even further reduction pos-sible. If these claims and other plant improve-ments and cost reductions are verified,advanced LWRs will be in a very good positionto drive a large share of the global growth sce-nario market.

For future LWR development, we recommendimplementation of designs that use a combina-tion of passive and active features in order to

it is important to maintain the principle thatthe primary responsibility for safe operation ofnuclear plants rests with the plant owners andoperators, as the generation segment of theelectric power industry is deregulated, and thatthe Nuclear Regulatory Commission shouldadapt its inspection activities, reportingrequirements, and enforcement actions toreflect the new incentives created by competi-tive generation markets.

REACTOR SAFETY

The global growth scenario considered in thisreport is a three-fold increase in the worldnuclear fleet capacity by 2050. The goal, ofcourse, should be to carry out this large expan-sion without increasing the frequency of seri-ous accidents. We believe this can be accom-plished by means of both evolutionary and newtechnologies focused on LWRs.

Three major reasons for reducing the fre-quency of serious accidents are: first, andforemost, they are a threat to public health.Reactor core damage has the potential torelease radioactivity to air and groundwater.Second, an accident destroys capital assets.Loss of a plant costs billions of dollars andcould restrict electrical generating capacity inthe locality until replacement, thereby addingto the economic loss. Third, a serious accidenterodes public confidence in nuclear genera-tion, with possible consequences of operatingplant shutdowns, and/or moratoria on newconstruction.

What is the expected frequency of accidentstoday with the currently operating nuclearplants? There are two ways to determine the fre-quency of accidents: historical experience andProbabilistic Risk Assessment.7 Since the begin-ning of commercial nuclear power in 1957,more than 100 LWR plants have been built andoperated in the U.S., with a total experience of2679 reactor-years through 2002. During thistime, there has been one reactor core damageaccident at Three Mile Island Unit 2. The core

MIT_ch06_47-52.qxd 7/16/2003 1:31 PM Page 48


enhance reliability of plant safety systems.Passive systems utilize stored energy for pump-ing, either by means of pressurized tanks or bygravity acting on water in elevated tanks. Theysubstitute for motor-driven pumps ultimatelydriven by emergency diesel generators, and canthereby remove the risk of failure of diesels tostart when needed, i.e., during a station black-out.

Additional gains may come with the introduc-tion of High-Temperature Gas Reactors(HTGRs). In principle the HTGR may be supe-rior to the LWR in its ability to retain fissionproducts in a loss-of- coolant accident, becauseof fuel form and because core temperatures canbe kept sufficiently low due to low power densi-ty design and high heat capacity of the core, ifRD&D validates this feature. Two HTGR plantsof small capacity and modular design are underdevelopment for eventual commercial applica-tion.

We describe briefly deployment for the globalgrowth scenario, first for LWRs, and then forHTGRs. Because of the experience base, con-struction of certified LWR designs at approvedsites could begin within the year or tworequired for contractual arrangements, limitedprimarily by retooling of a dormant industry,and obtaining regulatory approvals under newlicensing procedures. In order to build the glob-

years, an average rate of construction of 20 to 25plants10 per year would be required, withgreater numbers in later years. For historicalcomparison, LWR actual worldwide construc-tion totaled about 400 plants over 25 years, foran average of 16 plants completed per year.Doubling the past rate of construction for thisscenario is not an unreasonable projection, butremains a challenge, because plant constructiontime must also be reduced in order to reduceplant capital cost.

LWR experience does not exclude entry of theHTGR into the marketplace. However, it doesfocus attention on the lead times and costs asso-ciated with its development and the need for

operating experience before commitment ofcapital investment and the large manufacturingexpansion required to carry it out.

We believe that the lead time to carry outRD&D requirements for HTGR licensing, andat least several years of operation by one ormore demonstration plants, will add up to 15 to20 years before rapid, commercial deploymentcan be expected. Given this lead time, we expectthat two thirds or more of the fleet through2050 will be LWRs.

It is possible with success at every turn thatHTGR deployment could make up as much asone third of the global growth scenario. Theuncertainties in this projection are large, how-ever, and a range of HTGR penetration fromvery small to a high of one third is realistic. Wenote that the plant capacity of the two HTGRconcepts is in the range of 125-350 MWe, i.e.,substantially smaller than LWR plants. This isa very attractive feature of HTGRs, if cost tar-gets are met. Depending on the market sharesof the two HTGR concepts, about 4 plantswould be required to equal the output of a1000 MWe LWR. If HTGR plants were to cap-ture one third of the mid-century scenario,there would be about twice as many HTGRs asLWRs in 2050.

TRAINING AND QUALIFICATION OF PLANTMANAGEMENT AND STAFF

Realization of the mid-century scenario hasimportant implications for safety, and especial-ly in training and qualification of people com-petent to manage and operate the plants safely,including the supporting infrastructure neces-sary for maintenance, repair, refueling, andspent fuel management. Development of com-petent managers and identification of effectivemanagement processes is a critical element inachieving safe and economic nuclear powerplant operations. For developed countries thatnow operate nuclear plants, these tasks requireattention to the rejuvenation of the entire work-force.11

MIT_ch06_47-52.qxd 7/16/2003 1:31 PM Page 49

al growth scenario capacity of 1000 GWe in 50


For developing countries, however, this chal-lenge is much greater, because of the lack ofworkers in the many skills required in nuclearpower plant construction, operations, andmaintenance. The workforce must be trainedand grow from a small or negligible base. Thereare two main models for realization of the nec-essary growth: first, “do it yourself,” and second,the commercial mode of importing goods andservices. The first takes time and is subject toerror in the process of learning. The second isexpensive in the long run and fails to createskills and provide jobs at home. The best pathfor most developing countries is likely to besome combination of the two models that yieldsboth competence and jobs.

TERRORIST ATTACK ON NUCLEAR INSTALLATIONS

Terrorists have demonstrated their ability toinflict catastrophic damage. Nuclear facilities aspotential targets have not escaped notice. Onthe one hand experts have concluded that civilworks and security provisions make nuclearplants hard targets. On the other hand, the haz-ards are on a scale previously considered to beextremely rare in evaluation of severe reactoraccidents. The question is what new securitymeasures, if any, are appropriate? We believethere is no simple, one-size-fits-all answer. Itdepends on many factors including threat eval-uation, plant location, facility design, and gov-ernment security resources and practices.

Nuclear plant safety is a good starting point forthe evaluation of security risk. What we con-clude about plants also applies to other fuelcycle facilities. Nuclear plant safety has consid-ered natural external events, such as earth-quakes, tornadoes, floods, and hurricanes.Terrorist attack by fire or explosion is analogousto external natural events in its implication fordamage and release of radioactivity. Thestrength of containment buildings and struc-tures presents a major obstacle and hardenedtarget for attack. The Electric Power ResearchInstitute12 carried out an evaluation of aircraft

crash and NPP structural strength, concludingthat U.S. containments would not be breached.The U.S. NRC is performing its own evaluation,including structural testing at Sandia NationalLaboratory, not yet complete.

A broad survey and evaluation of hazards andprotective actions is in order to make decisionson adequate protection. Such a survey mustbegin by identifying possible modes of attackand vulnerabilities associated with designs andlocations. It must also identify the cost effective-ness of a range of security options for newdesigns, old plants near decommissioning, andplants in mid-life. There is also a need for shar-ing information with governments of countriesand supporting institutions that will undertakenuclear power programs in order to provideeffective intelligence and security.

NUCLEAR FUEL CYCLE SAFETY

Realization of the global growth scenario entailsconstruction and operation of many fuel cyclefacilities around the world, such as thosedescribed in Chapter IV, and also the facilitiesand repositories associated with waste manage-ment. There are varying degrees of risk to pub-lic safety associated with these facilities, andtherefore a need for systematic evaluation ofrisk on a consistent basis that takes into accountevaluations performed heretofore on individualfuel cycle facilities.

The need for such an evaluation is especiallyimportant in the case of reprocessing plants.The United States does not have any commer-cial reprocessing plants. France, the UnitedKingdom and Japan have reprocessing plants inoperation, based on aqueous PUREX separa-tions technology and improvements to it overmany years. Pyro-reprocessing and dry repro-cessing R&D has been done with no commer-cial application as yet. Aqueous separationplants have high inventories of fission products,as well as fissile material of work in process, andmany waste streams. Future improvements inseparation technology may be capable of reduc-

MIT_ch06_47-52.qxd 7/16/2003 1:31 PM Page 50


ing radioactive material inventories, measuredas a fraction of annual throughput, but invento-ries will continue to be large, because of thelarge annual product required, if and whenreprocessing comes into wider commercial usemany years in the future.

We are concerned about the safety of reprocess-ing plants,13 because of large radioactive mate-rial inventories, and because the record of acci-dents, such as the waste tank explosion atChelyabinsk in the FSU, the Hanford waste tankleakages in the United States and the dischargesto the environment at the Sellafield plant in theUnited Kingdom. Releases due to explosion orfire can be sudden and widespread. Althoughreleases due to leakage may take place slowly,they can have serious long-term public healthconsequences, if they are not promptly broughtunder control. Although the hazards of repro-cessing plants differ from those of reactors, theconcepts and methods and practices of reactorsafety are broadly applicable to assuring thesafety of reprocessing plants. We do not see theneed for commercial reprocessing in the globalgrowth scenario, but we believe the subjectrequires careful study,14 and action, if and whenreprocessing becomes necessary.

NOTES

1. Capacity factor is the ratio of actual annual plant electri-cal production and maximum annual production capa-bility.

2. While worldwide capacity factors (around 75%) arelower than those recently achieved in the U.S., a similartrend of improved capacity factors is observed outsideof the U.S. as well.

3. The Institute of Nuclear Power Operations in the U.S. and

4. Windscale, UK, gas-cooled reactor, graphite combustiondue to graphite stored heat release, with limited releaseof radioactivity, 1952; TMI 2, PWR, loss-of-coolant, 20%core meltdown, and small release, 1979; Chernobyl,graphite-moderated, water-cooled reactor, reactivityaccident with large external release of radioactivity andhealth effects, 1986.

5. Chelyabinsk, FSU, reprocessing waste explosion, (1957);Hanford, Washington State, waste storage tank leakage,(1970-); Sellafield, UK, reprocessing waste discharges intoocean, (1995-), Tokai-Mura, Japan, nuclear criticality inci-dent in fuel fabrication, (1999). We know of no completeinventory of reprocessing accidents; such a survey isneeded.

6. A similar event was discovered at a French nuclearpower plant in 1991.

7. Three important references are: Reactor Safety Study,WASH 1400, U.S. Nuclear Regulatory Commission,October 1975; Severe Accident Risks, NUREG-1150, U.S.NRC, December 1990; and Individual Plant ExaminationProgram, NUREG-1560, U.S. NRC, December 1997.

8. The number of core damage accidents expected is theproduct of the CDF and the reactor-years of experience.We assume a CDF of 10-4 and 40,000 reactor-years expe-rience during the period of 2005 to 2055: the product is4 accidents. The Safety Appendix 6 explains the relevantdata in more detail.

9. Potentially large release of radioactivity from fuel accom-panies core damage. Public health and safety dependson the ability of the reactor containment to preventleakage of radioactivity to the environment. If contain-ment fails, there would be a large, early release (LER) andexposure of people for some distance beyond the plantsite boundary, with the amount of exposure dependingon accident severity and weather conditions. The proba-bility of containment failure, given core damage, is about0.1. Hence the frequency of a LER is 1 in 1,000,000 years.LER is defined in U.S. NRC Regulatory Guide 1.174.

10. We expect individual plant capacities in the range of

smaller average capacity in developing countries.

11. The workforce has been aging for more than ten yearsdue to lack of new plant orders and decline of industrialactivity.

12. Deterring Terrorism - Aircraft Crash Impact AnalysesDemonstrate Nuclear Power Plant’s Structural Strength;EPRI Study, Nuclear Energy Institute website,www.nei.org, December 2002.

13. A brief comparison of reprocessing plants with reactorsshows that the historical accident frequency of repro-cessing plants is much larger than reactors: three of themore significant accidents are cited in footnote 5.Furthermore, the number of reprocessing plant-years ofoperation is many fewer that in the case of reactors.Therefore the accident frequency of reprocessing plantsis much higher.

14. We are not aware of PRA analyses of fuel cycle facilities;one exception is: Status report on the EPRI fuel cycleaccident risk assessment, prepared by SAIC for EPRIreport number NP-1128, July 1979.

MIT_ch06_47-52.qxd 7/16/2003 1:31 PM Page 51

600-1500 MWe. In developed countries the averageplant capacity is expected to be about 1000 MWe, with a

the World Association of Nuclear Operators worldwide.

MIT_ch06_47-52.qxd 7/16/2003 1:31 PM Page 52

53C h a p t e r 7 — S p e n t F u e l / H i g h - L e v e l Wa s t e M a n a g e m e n t

The management and disposal of radioactivewaste from the nuclear fuel cycle is one of themost difficult problems currently facing thenuclear power industry. Today, more than fortyyears after the first commercial nuclear powerplant entered service, no country has yet suc-ceeded in disposing of high-level nuclear waste– the longest-lived, most highly radioactive, andmost technologically challenging of the wastestreams generated by the nuclear industry.1

In most countries, the preferred technologicalapproach is to dispose of the waste in reposito-ries constructed in rock formations hundredsof meters below the earth’s surface. Althoughseveral experimental and pilot facilities havebeen built, there are no operating high-levelwaste repositories, and all countries haveencountered difficulties with their programs.The perceived lack of progress towards success-ful waste disposal clearly stands as one of theprimary obstacles to the expansion of nuclearpower around the world. 2

THE GOALS OF NUCLEAR WASTE MANAGEMENT AND DISPOSAL

Spent nuclear fuel discharged from nuclearreactors will remain highly radioactive formany thousands of years. The primary goal ofnuclear waste management is to ensure that thehealth risks of exposure to radiation from thismaterial are reduced to an acceptably low levelfor as long as it poses a significant hazard.Protection against the risk of malevolent inter-vention and misuse of the material is also nec-essary.

Because of the very long toxic lifetime of thewaste, the primary technical challenge is that oflong- term isolation. However, shorter-termrisks must also be addressed. Prior to final dis-position, the waste will pass through severalintermediate stages or operations, includingtemporary storage, transportation, condition-ing, packaging, and, potentially, intermediateprocessing and treatment steps. There are sever-al possible choices at each stage, and the designof the overall waste management system –including the specific technical characteristicsand the physical location of each stage – willimportantly affect the overall level of risk andits distribution over time. For example, wastemanagement strategies involving the separationof individual radionuclides from the spent fuelcould reduce long-term exposure risks, whileelevating risks in the short term. Such interde-pendencies attest to the importance of an inte-grated approach to nuclear waste managementdecision-making, in which the system-wideimpacts of individual decisions are fully consid-ered.

What constitutes an acceptable level of expo-sure risk? The U.S. Environmental ProtectionAgency (EPA) has stipulated that the radiationdose from all potential exposure pathways tothe maximally- exposed individual living closeto a waste disposal site should not exceed 15millirems per year for the first 10,000 years afterfinal disposition. This is about twenty times lessthan the dose that individuals receive annuallyfrom natural background radiation on average.EPA has translated the 15 millirem per yearstandard into an annual risk of developing afatal cancer of about 1 chance in 100,000.

Chapter 7 — Spent Fuel/High-Level Waste Management

MIT_ch07_53-64.qxd 7/16/2003 1:42 PM Page 53


We concur with the view that high-level wastecan safely be disposed of in geologic reposito-ries. As discussed below, we believe there areopportunities for advances in geologic andengineering system design that can provideadditional assurance regarding the long-termperformance of such repositories. We note,however, that among the general public, andeven among some in the technical community,there is a lack of confidence in the prospects forsuccessful technical and organizational imple-mentation of the geologic disposal concept.Previous missteps and failures in the wastemanagement programs of several countrieshave contributed to these doubts. Some mem-bers of the public – especially those living in thevicinity of proposed repository sites – alsoquestion the fairness and integrity of the siteselection process.

MEASURES TO INCREASE THE LIKELIHOOD OFSUCCESSFUL IMPLEMENTATION OF WASTEMANAGEMENT AND DISPOSAL

We have examined several possible innovationsthat might facilitate the successful implementa-tion of waste management and disposal. Inorder to make a difference, any such measureshould have to contribute significantly to oneor more of the following goals:

reduction of the risks to public health andsafety and the environment from waste man-agement and disposal activities in the shortand/or long term;

reduction of the economic costs of achievingan acceptable level of performance withrespect to short and long-term risk;

increase of public confidence in the technicaland organizational effectiveness of wastemanagement and disposal activities.

The innovations we have considered can begrouped into three categories:

technical modifications or improvementsthat could be incorporated into the once-through fuel cycle;

Different radiation exposure standards apply tooperating nuclear fuel cycle facilities.

The suitability of alternative waste manage-ment schemes must ultimately be judged inrelation to these fundamental safety goals.Other measures of waste management systemperformance are frequently cited, such as thevolume or mass of waste material generated, thetotal inventory of radioactivity in the waste, theamount of heat it emits, its radiotoxicity, andthe solubility and mobility of specific radionu-clides. Each of these metrics contains usefulinformation about the technical requirementsof individual components of the waste manage-ment system. But none of these metrics is anadequate proxy for the fundamental measure ofwaste management system performance — thatis, the risk to human health from radiationexposure in the short and long term.

THE FEASIBILITY OF GEOLOGIC DISPOSAL

As already noted, most countries with nuclearpower programs have stated their intention todispose of their high-level waste in minedrepositories, hundreds of meters below theearth’s surface. The concept of deep geologicdisposal has been studied extensively for severaldecades, and there is a high level of confidencewithin the expert scientific and technical com-munity that this approach is capable of safelyisolating the waste from the biosphere for aslong as it poses significant risks.3 This assess-ment is based on: (1) an understanding of theprocesses and events that could transportradionuclides from the repository to the bios-phere; (2) mathematical models which, whencombined with information about specific sitesand repository designs, enable the long-termenvironmental impact of repositories to bequantified; and (3) natural analog studies whichhelp to build confidence that the analyticalmodels can be reliably extrapolated to the verylong time-scales required for waste isolation.

MIT_ch07_53-64.qxd 7/16/2003 1:42 PM Page 54


technical modifications or improvementsrequiring a closed fuel cycle;

institutional or organizational innovations.

It is important to emphasize that each innova-tion must be evaluated in terms of its impact onthe entire waste management system, includingnot only final disposal but also pre-disposalprocessing, transportation, and storage opera-tions. In the following paragraphs we summa-rize our findings concerning each category ofinnovations. More detailed discussions can befound in Appendix 7.

TECHNICAL MODIFICATIONS OR IMPROVE-MENTS TO SPENT FUEL MANAGEMENT IN THEONCE-THROUGH FUEL CYCLE

Extended interim storage of spent fuelAlthough most spent fuel destined for directdisposal will in practice be stored above groundfor many years because of the protractedprocess of developing high level waste reposito-ries, storage arrangements so far have mostlybeen ad hoc and incremental. We believe that aperiod of several decades of interim storageshould be incorporated into the design of thespent fuel management system as an integralpart of the system architecture.4 Such a storagecapability would:

provide greater flexibility in the event ofdelays in repository development;

allow a deliberate approach to disposal andcreate opportunities to benefit from futureadvances in relevant science and technology;

provide greater logistical flexibility, with cen-tralized buffer storage capacity facilitatingthe balancing of short and long-term storagerequirements, and enabling the optimizationof logistics, pre- processing, and packagingoperations;

allow countries that want to keep open theoption to reprocess their spent fuel to do sowithout actually having to reprocess;

create additional flexibility in repositorydesign, since the spent fuel would be olderand cooler at the time of emplacement in therepository; and

potentially reduce the total number of repos-itories required.

At-reactor storage will be feasible for some spentfuel, even for several decades. For the remainder,centralized storage facilities will be required.Internationally, a network of safeguarded, wellprotected central storage facilities will also yieldimportant non-proliferation benefits (seeChapter 8). The siting of temporary storage facil-ities will likely be difficult. Although the techni-cal issues involved are more straightforward thanfor geologic repositories, the task of persuadingaffected communities to accept such facilitiesmay be no less challenging. Nevertheless, makingprovision for several decades of temporary spentfuel storage would make for a more robust wastemanagement system overall, and could be cost-effective too, if the result was to postpone theonset of major spending on repository construc-tion and operation.

High burnup fuel The burnup of spent fuel –the amount of energy that has been extractedfrom a unit of fuel at the time of its dischargefrom the reactor – is a design choice for reactoroperators. In the past, the burnup of LWR fuelaveraged about 33 MWD/kg. An increase to 100MWD/kg is within technical reach, and evengreater increases are potentially achievable.

Increasing the burnup to 100 MWD/kg wouldyield a threefold reduction in the volume ofspent fuel to be stored, conditioned, packaged,transported, and disposed of per unit of elec-tricity generated. The corresponding reductionin the required repository storage volumewould be more modest; the individual fuelassemblies, although there would be fewer ofthem, would generate more decay heat andwould therefore have to be spaced farther apartin the repository. The amount of plutoniumand other actinides, which are the dominantcontributors to the radiotoxicity of the spentfuel after the first hundred years or so, would

MIT_ch07_53-64.qxd 7/16/2003 1:42 PM Page 55


also be reduced somewhat per unit of electrici-ty generated. A further benefit of higher burnupis that the isotopic composition of the dis-charged plutonium would make it less suitablefor use in nuclear explosives.5

It is important to note, however, that the pres-ent pricing structure for nuclear waste manage-ment services in the United States – a standardfee of one-tenth of a cent payable to the govern-ment on each kilowatt hour of nuclear electric-ity generated — provides no economic incen-tive for nuclear generators to move in the direc-tion of higher burnup. No discount is providedfor the reduced volume of spent fuel and thesafety, proliferation resistance, and economicbenefits associated with higher burnup.6

Advances in geologic repository design A geo-logic repository must provide protectionagainst every plausible scenario in whichradionuclides might reach the biosphere andexpose the human population to dangerousdoses of radiation. Of all possible pathways, theone receiving most attention involves ground-water seeping into the repository, the corrosionof the waste containers, the leaching ofradionuclides into the groundwater, and themigration of the contaminated groundwatertowards locations where it might be used asdrinking water or for agricultural purposes.Although the details differ, all proposed reposi-tory designs adopt a ‘defense in depth’ approachto protecting against this scenario, relying on acombination of engineered components andnatural geologic, hydrologic, and geochemicalbarriers to contain the radionuclides.

The engineered barriers, broadly defined toinclude those physical and chemical features ofthe near- field environment that affect the con-tainment behavior of the waste packages, havean important role to play in the overall per-formance of the repository. To date there hasnot been an adequate technical basis for theselection and development of the engineeredbarriers in the context of the overall multi- bar-rier system.

In siting a repository, it is important to select ageochemical and hydrological environment thatwill ensure the lowest possible solubility andmobility of the waste radionuclides. The geo-chemical conditions in the repository host rockand surrounding environment strongly affectradionuclide transport behavior. For example,several long-lived radionuclides that are poten-tially important contributors to long-term dose,including technetium-99 and neptunium-237,are orders of magnitude less soluble in ground-water in reducing environments than underoxidizing conditions.

Alternative disposal technologies: The deepborehole approach An alternative to buildinggeologic repositories a few hundred metersbelow the earth’s surface is to place waste canis-ters in boreholes drilled into stable crystallinerock several kilometers deep. Canisters contain-ing spent fuel or high-level waste would be low-ered into the bottom section of the borehole,and the upper section – several hundred metersor more in height – would be filled with sealantmaterials such as clay, asphalt, or concrete. Atdepths of several kilometers, vast areas of crys-talline basement rock are known to be extreme-ly stable, having experienced no tectonic, vol-canic or seismic activity for billions of years.

The main advantages of the deep borehole con-cept relative to mined geologic repositoriesinclude: (a) a much longer migration pathwayfrom the waste location to the biosphere; (b) thelow water content, low porosity and low perme-ability of crystalline rock at multi-kilometerdepths; (c) the typically very high salinity of anywater that is present (because of its higher den-sity, the saline water could not rise convectivelyinto an overlying layer of fresh water even ifheated); and (d) the ubiquity of potentially suit-able sites.

An initial screening suggests that most of thecountries that are likely to employ nuclearpower in our global growth scenario may havegeology appropriate for deep waste boreholes.Co-location of boreholes with reactor sites is apossibility. Suitable host rock also occurs

MIT_ch07_53-64.qxd 7/16/2003 1:42 PM Page 56


beneath the sea floor. For this reason the con-cept may be particularly interesting for denselypopulated countries like Japan, Korea, andTaiwan. Since most of the power reactors inthese countries (and indeed in most countries)are located on or close to the coast, the possibil-ity arises of constructing artificial offshoreislands which would be ideal sites from which todrill beneath the seabed and which could alsoserve as temporary storage venues for the spentfuel, obviating the need for on-land waste trans-portation and storage.

The overall system cost of deep borehole dis-posal using conventional drilling technology isuncertain, but according to one estimate wouldbe comparable to that of mined geologic dis-posal.7 Advances in technology could reduce thecost of drilling significantly. But since drillingalone accounts for only a relatively small frac-tion of the overall costs, the opportunities forsavings are limited. A more important econom-ic advantage may derive from the modularity ofthe deep borehole concept and the more flexiblesiting strategy that it allows. 8

Implementing the deep borehole scheme wouldrequire the development of a new set of stan-dards and regulations, a time-consuming andcostly process. A major consideration would bethe difficulty of retrieving waste from boreholesif a problem should develop (though the greaterdifficulty of recovering the plutonium in thewaste might also be an advantage of the bore-hole scheme). Current U.S. regulatory guide-lines for mined repositories require a period ofseveral decades during which the high levelwaste should be retrievable. This would be dif-ficult and expensive to ensure in the case ofdeep boreholes, though probably not impossi-ble. Moreover, at the great depths involved,knowledge of in situ conditions (e.g., geochem-istry, stress distributions, fracturing, water flow,and the corrosion behavior of different materi-als) will never be as comprehensive as in shal-lower mined repository environments.Recovery from accidents occurring duringwaste emplacement – for example, stuck canis-ters, or a collapse of the borehole wall – is also

likely to be more difficult than for correspon-ding events in mined repositories. Finally,despite the order of magnitude increase in thedepth of waste emplacement, it is difficult topredict the impact on public opinion of a shiftin siting strategy from one large central reposi-tory to scores of widely dispersed boreholes.

Despite these obstacles, we view the deep boreholedisposal approach as a promising extension ofgeological disposal, with greater siting flexibilityand the potential to reduce the already very lowrisk of long-term radiation exposure to still lowerlevels without incurring significant additionalcosts.

TECHNICAL MODIFICATIONS REQUIRING ACLOSED FUEL CYCLE

We next consider a set of waste managementoptions involving the extraction of radionu-clides from the spent fuel. The motivations forwaste separation can be inferred from Figures7.1, 7.2, and 7.3. At different times, differentradionuclides are the dominant contributors tooverall radioactivity and radiotoxicity and tothe radioactive decay heat emitted by the fuel.Partitioning the spent fuel into separateradionuclide fractions and managing each frac-tion according to its particular characteristicscould create additional flexibilities and newopportunities to optimize the overall wastemanagement system. Partitioning also createsthe opportunity to transmute the most trouble-some radionuclides into more benign species.Thermal reactors, fast reactors, and acceleratorshave all been investigated as candidate transmu-tation devices, both individually and in combi-nation.

Decisions about partitioning and transmuta-tion must also consider the incremental eco-nomic costs and safety, environmental, and pro-liferation risks of introducing the additionalfuel cycle stages and facilities necessary for thetask.9 These activities will be a source of addi-tional risk to those working in the plants, as wellas the general public, and will also generate con-

MIT_ch07_53-64.qxd 7/16/2003 1:42 PM Page 57


siderable volumes of non-high- level waste con-taminated with significant quantities oftransuranics. Much of this waste, because of itslong toxic lifetime, will ultimately need to bedisposed of in high-level waste repositories.Moreover, even the most economical partition-ing and transmutation schemes are likely to addsignificantly to the cost of the once-throughfuel cycle.10

We first consider the option of waste partition-ing alone, and then the combination of parti-tioning and transmutation.

Waste partitioning Two fission products,strontium-90 and cesium-137, each with half-lives of about 30 years, account for the bulk ofthe radioactivity and decay heat in spent fuel

starting a few years after discharge and for thenext several decades. Thereafter, the actinides asa group become the dominant contributors todecay heat and radiotoxicity, with differentactinides dominating at different times.

Extracting the high-heat-emitting fission prod-uct radionuclides from the spent fuel and stor-ing them separately would allow the remainderof the radionuclides to occupy a more compactvolume in a geologic repository, perhaps evenreducing the total number of repositoriesrequired. It should be noted, however, that asimilar result could be achieved without theneed for separation by storing the spent fuel forseveral decades to allow the fission products todecay. In this case, moreover, there would be noneed for a separate storage facility for the parti-tioned strontium-90 and cesium-137, whichwould have to be isolated from the biospherefor several hundred years before radioactivedecay would render them harmless.

An alternative strategy would be to partition theuranium, plutonium and the other actinidesfrom the spent fuel. If actinide partitioningwere implemented in conjunction with interimwaste storage for long enough to allow thestrontium-90 and cesium-137 to decay signifi-cantly before repository emplacement, theeffective storage capacity of a given repositorycould be increased many-fold. But the parti-tioned actinides would still have to be stored ina separate repository (or alternatively in deepboreholes). Moreover, by separating theactinides from the more radioactive fissionproducts, the radiation barrier against unau-thorized recovery of weapons-usable plutoni-um would be reduced relative to the case ofintact spent fuel, at least for a century or so.

The case for partitioning the spent fuel and sep-arately storing the different radionuclide frac-tions does not seem persuasive, especially giventhe additional costs and near-term environ-mental and safety risks associated with parti-tioning operations.

Figure 7.1 Radioactivity profile of spent fuel (curies/MTHM)

Basis: PWR Spent Fuel 50 MWd/kg HM 4.5% initial enrichment

MIT_ch07_53-64.qxd 7/16/2003 1:42 PM Page 58


Waste partitioning and transmutation Wastepartitioning strategies potentially become moreattractive when combined with transmutation.There are three principal motivations for parti-tioning/transmutation schemes. First, if thelong-lived isotopes in the waste could beextracted and destroyed, many more locationsmight become suitable candidates to host arepository for the remaining material. Indeed, ifall of the long-lived radionuclides could beremoved and destroyed, a disposal strategy rely-ing solely on engineered structures for radionu-clide containment might become feasible. Theactinides, which as a group dominate theradiotoxicity of the spent fuel after about 100years (see Figure 7.3), are usually cited as theprime candidates for partitioning and transmu-tation. However, performance assessments ofthe proposed repository sites at YuccaMountain and at Olkiluoto in Finland showthat long-lived fission products, such as tech-netium-99 and iodine-129, are more importantthan most actinides as sources of long-termexposure risk.11 Partitioning and transmutationstudies have yet to show that these fission prod-ucts can be dealt with effectively. Even for theactinides, the technology is not yet available toremove these isotopes from all fuel cycle wastestreams, and complete elimination of these iso-topes from secondary, as well as primary wastestreams, is unlikely ever to be attractive on eco-nomic grounds.

A second motivation for partitioning and trans-mutation is to reduce the thermal load on therepository, thereby increasing its storage capac-ity. As Figure 7.2 shows, after 60–70 years, theactinides are the dominant contributors towaste heating. As previously noted, actinidepartitioning and transmutation, combined witha period of several decades of interim storageprior to final disposal of the residual waste,could increase the effective storage capacity of agiven repository several-fold. Given the extremedifficulty of repository siting in most countries,any reduction in the required number of repos-itories must be counted as a significant gain,although this would be at least partly offset bythe additional difficulty of siting the necessary

waste partitioning and related fuel cycle facili-ties. As noted above, a less costly way to increasethe effective storage capacity of repositorieswould simply be to defer waste emplacementuntil more of the heat-emitting radionuclideshave decayed. In some countries, moreover,especially those with relatively small nuclearprograms, a single repository is likely to be ableto accommodate the entire national inventoryof high-level waste even without actinide parti-tioning. 12

A third motivation for partitioning and trans-mutation is to eliminate the risk that plutoniumcould later be recovered from a repository andused for weapons. It is difficult to assess the sig-nificance of this result. The value today of elim-

Figure 7.2 Decay Heat Profile of Spent Fuel


MIT_ch07_53-64.qxd 7/16/2003 1:42 PM Page 59


inating the technical means for one particulartype of aggressive or malevolent human behav-ior centuries or millennia from now, out of allpossible opportunities for such behavior thatmay exist at that time, is a question perhaps bet-ter addressed by philosophers than engineers,political scientists, or economists. From a nar-rowly technical perspective the best that can besaid is that, without partitioning and transmu-tation, the feasibility of plutonium recoveryfrom a repository will increase with time, as theradiation barrier created by the fission productsin the waste decays away.

Against these putative long-term benefits ofwaste partitioning and transmutation must beweighed the increased short-term health, safety,

environmental, and security risks involved. Allactinide partitioning and transmutationschemes currently under consideration alsoseem likely to add significantly to the economiccost of the nuclear fuel cycle.

The trade-off between reduced risk over verylong time scales and increased risk and cost inthe short term is an issue on which reasonablepeople can disagree. The evaluation can fur-thermore be expected to vary by country,reflecting the different preferences and differentconstraints – geological, demographic, political,economic – of different societies. Nevertheless,taking all these factors into account, we do notbelieve that a convincing case can be made on thebasis of waste management considerations alonethat the benefits of advanced fuel cycle schemesfeaturing waste partitioning and transmutationwill outweigh the attendant risks and costs.Future technology developments could changethe balance of expected costs, risks, and bene-fits. For our fundamental conclusion to change,however, not only would the expected long-term risks from geologic repositories have to besignificantly higher than those indicated in cur-rent risk assessments, but the incremental costsand short-term safety and environmental riskswould have to be greatly reduced relative to cur-rent expectations and experience.

Some argue that partitioning and transmuta-tion, by reducing the toxic lifetime of the waste,could change public attitudes towards the feasi-bility and acceptability of nuclear waste dispos-al. There is no empirical evidence of which weare aware to support this view. Our own judg-ment is that local opposition to waste reposito-ries or waste transportation routes would not bemuch influenced, even if the toxic lifetime werereduced from hundreds of thousands to hun-dreds of years.

Our assessment of alternative waste manage-ment strategies leads to the following importantconclusion: technical improvements to the wastemanagement strategies in the once-through fuelcycle are potentially available that could yield ben-efits at least as large as those claimed for advanced

Figure 7.3 Radiotoxicity Index for 1MT of Spent Fuel


MIT_ch07_53-64.qxd 7/16/2003 1:42 PM Page 60


fuel cycles featuring waste partitioning and trans-mutation, and with fewer short-term risks. Themost that can reasonably be expected of parti-tioning and transmutation schemes is to reducethe inventory of actinides in geologic reposito-ries by perhaps two orders of magnitude.13

Reductions of two orders of magnitude or morein long-term radiation exposure risks couldpotentially be achieved by siting the repositoriesin host environments in which chemicallyreducing conditions could be ensured.Moreover, deep borehole technology offers acredible prospect of risk reductions of severalorders of magnitude relative to mined reposito-ries. Neither of these options is likely to cost asmuch or take as long to develop and deploy aswaste partitioning and transmutation schemes.

INSTITUTIONAL INNOVATIONS

Technological advances can increase the likeli-hood that nuclear waste disposal will be suc-cessfully implemented. But an equally impor-tant consideration is the competence of theimplementing authorities. A major challengefor these authorities under our global growthscenario will be to find suitable disposal sites. Aworldwide deployment of one thousand 1000megawatt LWRs operating on the once-throughfuel-cycle with today’s fuel management char-acteristics would generate roughly three timesas much spent fuel annually as does today’snuclear power plant fleet.14 If this fuel was dis-posed of directly, new repository storage capac-ity equal to the currently planned capacity ofthe Yucca Mountain facility would have to becreated somewhere in the world roughly everythree or four years. For the United States, athree-fold increase in nuclear generating capac-ity would create a requirement for a YuccaMountain equivalent of storage capacity rough-ly every 12 years (or every 25 years if the physi-cal rather than the legal capacity limit of YuccaMountain is assumed.) Even if the technicalstrategies discussed above succeed in reducingthe demand for repository capacity, the organi-zational and political challenges of siting willsurely be formidable.

Today the political and legal mechanisms forbalancing broad national policy goals againstthe concerns of affected local communities inthe site selection process vary widely, evenamong the democratic societies of the West.This diversity of approaches will surely persist,although over time, as some nations achievesuccess in gaining local acceptance of reposito-ries, some international diffusion of ‘best sit-ing practices’ is probable. On present evidence,these best practices seem likely to include fullaccess to information, opportunities forbroad-based and continuing local communityparticipation in consensus-building processes,the adoption of realistic and flexible schedules,and a willingness not merely to compensatelocal communities for hosting facilities, butalso to find ways to make them actually betteroff.

Another important requirement for successfulwaste management implementation is the effec-tive administration of a large-scale industrialoperation involving the transportation, storage,processing, packaging, and emplacement oflarge quantities of radioactive waste. In theUnited States, as a matter of law and policy, thegovernance and management structure of thehigh-level waste program has been heavilyfocused on the development of the YuccaMountain project. The scientific and engineer-ing effort has also been almost exclusivelyfocused on the investigation of the YuccaMountain site and the development of a repos-itory design for that site. However, the organiza-tional and managerial demands of repositorysiting – a one-time project that is by definitionexploratory, developmental, and, inevitably,highly politicized – are fundamentally differentfrom the demands of a routine-based large-scale industrial processing and logistics opera-tion. The intense focus on the Yucca Mountainproject will continue as design and licensingactivities gain momentum over the next fewyears. In addition, the U.S. high level waste man-agement program will require (1) a broadly-based, long-term R&D program, and (2) a sepa-rate organization for managing the operations ofthe waste management system.

MIT_ch07_53-64.qxd 7/16/2003 1:42 PM Page 61


Finally, we note that international cooperationin the field of high-level waste management anddisposal is presently under-developed. Strongerinternational coordination of standards andregulations for waste transportation, storage,and disposal will be necessary in order tostrengthen public confidence in the safety ofthese activities. There is also considerablepotential for international sharing of wastestorage and disposal facilities. This might notonly reduce proliferation risks from the fuelcycle (as discussed in the following chapter),but could also yield significant economic andsafety benefits, although formidable politicalobstacles will have to be overcome first.

The authors of this study wish to acknowledgethe valuable research support provided by ourformer students, Dr. Brett Mattingly and Dr. David Freed in the preparation of this chapter.

NOTES

1. In this study we focus on spent fuel and reprocessedhigh-level waste, since these waste types contain mostof the radioactivity generated in the nuclear power fuelcycle and pose the greatest technical and political chal-lenges for final disposal. We also include in the discus-sion so-called TRU waste — non-high-level waste con-taminated with significant quantities of long-livedtransuranic radionuclides — which because of itslongevity will likely be disposed of in the same facilitiesas high-level waste. Other types of nuclear waste, includ-ing low-level waste and uranium mill tailings, are gener-ated in larger volumes in the nuclear fuel cycle but posefewer technical challenges for disposal, although local-ized opposition to disposal facilities for these materialshas sometimes been intense.

2. In the opinion survey commissioned for this study,almost two-thirds of respondents did not believe thatnuclear waste could be safely stored for long periods.

3. According to one recent international scientific assess-ment,“[I]n a generic way, it can be stated with confi-dence that deep geologic disposal is technically feasibleand does not present any particularly novel rock engi-neering issues. The existence of numerous potentiallysuitable repository sites in a variety of host rocks is alsowell established.” (International Atomic Energy Agency,“Scientific and Technical Basis for the Geologic Disposalof Radioactive Wastes”, Technical Report No. 413, IAEA,Vienna, 2003.) Another expert group, convened by theOECD’s Nuclear Energy Agency, found that,“[T]here istoday a broad international consensus on the technicalmerits of the disposal of long-lived radioactive waste indeep and stable geologic formations…. Currently, geo-logic disposal can be shown to have the potential toprovide the required level and duration of isolation.”“The Environmental and Ethical Basis of GeologicDisposal of Long-Lived Radioactive Wastes: A CollectiveOpinion of the Radioactive Waste ManagementCommittee of the OECD Nuclear Energy Agency,”1995 athttp://www.nea.fr/html/rwm/reports/1995/geodisp.html. Yet another recent international assessment, this timeunder the auspices of the U.S. National Academy ofSciences, found that,“geological disposal remains theonly scientifically and technically credible long-termsolution available to meet the need for safety withoutreliance on active management…a well-designed repos-itory represents, after closure, a passive system contain-ing a succession of robust safety barriers. Our presentcivilization designs, builds, and lives with technologicalfacilities of much greater complexity and higher hazardpotential.” See National Academy of Sciences,Board onRadioactive Waste Management, Disposition of HighLevel Waste and Spent Nuclear Fuel: The ContinuingSocietal and Technical Challenges, National AcademyPress, Washington, D.C., 2001.

4. Because of the high heat generation, spent fuel must bestored for at least five years before it can be emplaced ina geologic repository. After another 30 years, the decayheat from the fission products Cs-137 and Sr-90, theleading sources of heat during this period, will havehalved. After 100 years, the contribution from these iso-topes will have declined by more than 90%. At thatpoint, the fission product radiation barrier, which untilthen would complicate attempts by would-be prolifera-tors to recover plutonium from the spent fuel, will havelargely dissipated, and storage in relatively accessiblesurface or near-surface facilities thereafter would be lessdesirable on non-proliferation grounds.

MIT_ch07_53-64.qxd 7/16/2003 1:42 PM Page 62


5. As the burnup increases, the proportion of plutonium-239 in the plutonium declines, while the proportion ofPu-238 increases. For example, an increase in the bur-nup of PWR fuel from 33 MWD/kg to 100 MWD/kgwould result in a decline in the Pu-239 content from65% to 53%, while the Pu-238 content would increasefrom 1% to about 7%. (Zhiwen Xu, Ph.D. dissertation,Department of Nuclear Engineering, M.I.T., 2003). Pu-238is a particularly undesirable isotope in nuclear explosivesbecause of its relatively high emission rate of sponta-neous fission neutrons and decay heat. According tosome specialists, a Pu- 238 content above about 6%would make plutonium essentially unusable forweapons purposes. The denaturing effect of Pu-238would be limited to a couple of centuries, however,because of its relatively short (87-year) half-life.

6. In recent years the average burnup of LWR fuel has risenfrom about 33 MWD/kg to about 45–50 MWD/kg. LWRoperators have taken this step for economic reasonsthat are largely unrelated to waste disposal; the higher-burnup fuel cycle allows the reactors to operate forlonger periods between refueling, thus increasing thereactor capacity factor.

7. Weng-Sheng Kuo, Michael J. Driscoll, and Jefferson W.Tester,“Re-evaluation of the deep drillhole concept fordisposing of high-level nuclear wastes,” Nuclear ScienceJournal, vol. 32, no. 3, pp. 229–248, June 1995.

8. According to one recent estimate, a full-scale 4-kilome-ter deep borehole could be drilled and cased in lessthan 5 months, at a cost of about $5 million. TimHarrison,“Very Deep Borehole: Deutag’s Opinion onBoring, Canister Emplacement and Retrievability”,Swedish Nuclear Fuel and Waste Management Co., R- 00-35, May 2000.

9. See, for example, National Academy of Sciences, NuclearWastes: Technologies for Separation and Transmutation,Committee on Separations Technology andTransmutation Systems, National Research Council,Washington, D.C., 1996; B.Brogli and R. A. Krakowski,

Paul Scherrer Institut Nuclear Energy and SafetyResearch Department, PSI Bericht No. 02-14, August2002.

10. The PUREX/MOX fuel cycle currently practiced in severalcountries is one variant of the waste partitioning/trans-mutation option, in which uranium and plutonium iso-topes are partitioned from the spent fuel, and the sepa-rated plutonium isotopes are partially transmuted intoshorter-lived fission products in light water reactors. Asshown in Appendix 5D, PUREX/MOX increases the fuelcycle cost to 4.5 times the once-through fuel cycle cost,depending on various assumptions.

11. To determine which radionuclides should be the princi-pal targets of partitioning and transmutation, it is neces-sary to assess the likelihood that individual radionu-clides will be transported from the repository to thebiosphere. This in turn is a function of the particular geo-chemical and hydrological characteristics of the reposi-tory environment. In the oxidizing conditions character-istic of Yucca Mountain, the dominant contributors tolong-term exposure risk are neptunium-237 and tech-netium-99. During the first 70,000 years, technetium-99is the leading contributor, and between 100,000 yearsand 1 million years, the dominant isotope is Np-237. Thepeak dose of about 150 millirems/year (about half thebackground dose) occurs after about 400,000 years. (See:Final Environmental Impact Statement for YuccaMountain Repository, February 2002) In contrast, a per-formance assessment of the proposed Finnish repositoryat Olkiluoto, in crystalline rock in a chemically reducingenvironment, concludes that the actinides would con-tribute very little to long-term dose, and that the domi-nant contributors would be a few long-lived fissionproducts. The projected peak dose, moreover, is threeorders of magnitude lower than that at Yucca Mountain(see Vieno and Nordman,“Safety Assessment of SpentFuel Disposal in Hastholmen, Kivetty, Olkiluoto andRomuvaara - TILA-99,” POSIVA 99-07, March 1999, ISBN951-652-062-6).

12. For the repository at Yucca Mountain, operating in theso-called higher-temperature operating mode, the totalsubsurface area that would be required to accommo-date the legal limit of 70,000 MT of spent fuel equivalent(including 7000 MT of defense high level waste) wouldbe 1150 acres, equivalent to a square roughly 2 kilome-ters along a side. U.S. Department of Energy,“YuccaMountain Science and Engineering Report, Rev. 1”,DOE/RW-0539-1, February 2002, Executive Summary, athttp://www.ymp.gov/documents/ser_b/. The currentfleet of U.S. reactors is expected to discharge at least105,000 MT of spent fuel and possibly considerablymore, depending on reactor operating lifetimes. The70,000 MTHM capacity limit at Yucca Mountain waspolitically determined, and according to some knowl-edgeable observers the physical storage capability ofthe site would be at least twice as large.

13. Nuclear Energy Agency, Accelerator-Driven Systems andFast Reactors in Advanced Fuel Cycles: A ComparativeStudy, OECD, 2002 (available athttp://www.nea.fr/html/ndd/reports/2002/nea3109.htm.

14. If each reactor has a burn-up of 50,000 MWth-d/MTHM ,

deployment of 1000 1 Gwe reactors would result in anannual spent fuel discharge of about 20,000 metric tonsper year.

MIT_ch07_53-64.qxd 7/16/2003 1:42 PM Page 63

“Degree of sustainability of various nuclear fuel cycles,”

a capacity factor of 0.9, and a thermal efficiency of 33%,

MIT_ch07_53-64.qxd 7/16/2003 1:42 PM Page 64

65C h a p t e r 8 — N o n p r o l i f e r a t i o n

Nuclear weapons proliferation has been promi-nent in discussions about nuclear power sinceits earliest days. The birth of nuclear technolo-gy that began with production of the firstweapons-usable fissionable material — pluto-nium production in nuclear reactors and high-enriched uranium by isotope enrichment —assured that this would be so. Today, the objec-tive is to minimize the proliferation risks ofnuclear fuel cycle operation. We must prevent theacquisition of weapons-usable material, eitherby diversion (in the case of plutonium) or bymisuse of fuel cycle facilities (including relatedfacilities, such as research reactors or hot cells)and control, to the extent possible, the know-how about how to produce and process eitherHEU (enrichment technology) or plutonium.

This proliferation concern has led, over the lasthalf century, to an elaborate set of internation-al institutions and agreements, none of whichhave proved entirely satisfactory. The NuclearNonproliferation Treaty (NPT) is the founda-tion of the control regime, since it embodies therenunciation of nuclear weapons by all signato-ries except for the declared nuclear weaponsstates – the P-5 (the United States, Russia, theUnited Kingdom, France, China) — and a com-mitment to collaborate on developing peacefuluses of nuclear energy. However, non-signato-ries India and Pakistan tested nuclear weaponsin 1998, and signatories, such as South Africaand North Korea, have admitted to makingnuclear weapons.

The International Atomic Energy Agency(IAEA) has responsibility for verifying NPTcompliance with respect to fuel cycle facilitiesthrough its negotiated safeguards agreements

with NPT signatories. The IAEA’s safeguardefforts, however, are seriously constrained bythe scope of their authorities (as evidenced inIraq, Iran, and North Korea during the lastdecade), by their allocation of resources, and bythe growing divergence between responsibilitiesand funding. The United Nations SecurityCouncil has not yet established a procedure orshown a willingness to impose sanctions whenIAEA safeguards agreements are violated. Avariety of multilateral agreements, such as theNuclear Supplier Group guidelines for exportcontrol, aim to restrict the spread of prolifera-tion-enabling nuclear and dual-use technology.European centrifuge enrichment technology,however, is known to have contributed toweapons development elsewhere, and the USand Russia have a continuing dispute overtransfer of Russian fuel cycle technologies toIran (a NPT signatory). This is not to say thatthe safeguards regime has failed to restrain thespread of nuclear weapons; it almost certainlyhas. Nevertheless, its shortcomings raise signif-icant questions about the wisdom of a globalgrowth scenario that envisions a major increasein the scale and geographical distribution ofnuclear power.

In addition to the risk of nuclear weapons capa-bility spreading to other nations, the threat ofacquisition of a crude nuclear explosive by asub-national group has arisen in the aftermathof the September 11, 2001 terrorist attacks. Thereport of interest in nuclear devices by the ter-rorist Al Qaeda network especially highlightsthis risk. Terrorist or organized crime groupsare not expected to be able to produce nuclearweapons material themselves; the concern istheir direct acquisition of nuclear materials by

Chapter 8 — Nonproliferation

MIT_ch08_65-70.qxd 7/16/2003 1:44 PM Page 65


acteristic of this scenario is that much of thedeployment would be expected in industrial-ized countries that either already have nuclearweapons, thus making materials securityagainst theft the principal issue, or are viewedtoday as minimal proliferation risks. The con-cern about these nations’ ability to providesecurity for nuclear material is especially elevat-ed for Russia, whose economic difficulties havelimited its effort to adopt strong material secu-rity measures; the concern applies to materialsfrom both the weapons program and the fuelcycle,1 which have significant inventories ofseparated Pu. Moreover geopolitical change, forexample, in East Asia, could change the interestsof some nations in acquiring nuclear capability.Japan, South Korea, and Taiwan have advancednuclear technology infrastructures and overseveral decades might adjust to the emergenceof China as both a nuclear weapons state and aregionally dominant economic force by seekingnuclear capability. North Korea provides a fur-ther complication to this dynamic.

The developing world might plausibly accountfor about a third of deployed nuclear power inthe mid-century scenario. An appreciable partof this will likely be in China and India, whichalready have nuclear weapons and dedicatedstockpile facilities and thus are not viewed asthe highest risks for fuel cycle diversion.Nevertheless, dramatic growth of nuclearpower in the sub-continent could be a pathwayfor nuclear arsenal expansion in India andPakistan. The security of their nuclear enter-prises remains of concern.

On the other hand, a number of other nationswith relatively little nuclear infrastructuretoday, such as the Southeast Asian countriesIndonesia, Philippines, Vietnam, and Thailand(with a 2050 projected combined populationover 600 million) are also likely candidates fornuclear power in the global growth scenario.Iran is actively pursuing nuclear power, withRussian assistance, even though it has vastunexploited reserves of natural gas and couldclearly meet its electricity needs more econom-ically and rapidly by using this domestic

theft or through a state sponsor. This places thespotlight on the PUREX/MOX fuel cycle as cur-rently practiced in several countries, since thefuel cycle produces during conventional opera-tion nuclear material that is easily made usablefor a weapon. The sub-national theft risk wouldbe exacerbated by the spread of thePUREX/MOX fuel cycle, particularly to thosecountries without the infrastructure for assur-ing stringent control and accountability.

A separate concern is the dirty bomb threat inwhich radioactive material (from any source,such as nuclear spent fuel or cobalt sourcesused in medicine and industry) is dispersed in aconventional explosive as a weapon of mass dis-ruption. The dirty bomb threat is a very serioussecurity concern but is not specific to thenuclear fuel cycle and will not be discussed fur-ther in the proliferation context.

It is useful to set a scale for the proliferation riskthat has emerged from nuclear power operationto date. Spent fuel discharged from power reac-tors worldwide contains well over 1000 tonnesof plutonium. While the plutonium is protect-ed by the intense radioactivity of the spent fuel,the PUREX chemical process most commonlyused to separate the plutonium with high puri-ty, is well known and described in the open lit-erature. With modest nuclear infrastructure,any nation could carry out the separation at thescale needed to acquire material for severalweapons. Further, the MOX fuel cycle has led toan accumulation of about 200 tonnes of sepa-rated plutonium in several European countries,Russia and Japan. This is equivalent to 25,000weapons using the IAEA definition of 8kg/weapon. Separated plutonium is especiallyattractive for theft or diversion and is fairly eas-ily convertible to weapons use, including bythose sub-national groups that have significanttechnical and financial resources.

The nonproliferation issues arising from theglobal growth scenario are brought into sharpfocus by examining a plausible scenario for thedeployment of 1000 GWe nuclear capacity (seeTable 3.2 and Appendix 2). An important char-

MIT_ch08_65-70.qxd 7/16/2003 1:44 PM Page 66


resource. The United States in particular hasargued that this indicates Iranian interest inacquiring a nuclear weapons capability, eventhough Iran is an NPT signatory and has a safe-guards agreement with the IAEA in place.Recent revelation of the spread of clandestinecentrifuge enrichment and heavy water technol-ogy exacerbates this concern. Thus the U.S. isarguing that cooperation with Iran on nuclearpower should cease irrespective of the NPT’scall for cooperation in the peaceful use ofnuclear energy (Article IV). This issue has beena significant irritant in U.S.-Russia relations.Such conflicts between an underlying principleof the NPT and the aims of specific countriescould become more common in the growth sce-nario.

The rapid global spread of industrial capacity(such as chemicals, robotic manufacturing) andof new technologies (such as advanced materi-als, computer-based design and simulationtools, medical isotope separation) will increas-ingly facilitate proliferation in developing coun-tries that have nuclear weapons ambitions. Afuel cycle infrastructure makes easier both theactivity itself and the disguising of this activity.Indeed, even an extensive nuclear fuel cycleRD&D program and associated facilities couldopen up significant proliferation pathways wellbefore commercial deployment of new tech-nologies.

We conclude that the current non-proliferationregime must be strengthened by both technicaland institutional measures with particular atten-tion to the connection between fuel cycle technol-ogy and safeguardability. Indeed, if the nonpro-liferation regime is not strengthened, the optionof significant global expansion of nuclear powermay be impossible, as various governmentsreact to real or potential threat of nuclearweapons proliferation facilitated by fuel cycledevelopment. The U.S. in particular should re-commit itself to strengthening the IAEA and theNPT regime.

The specific technical and institutional meas-ures called for will depend upon the fuel cycle

technologies that account for growth in theglobal growth scenario. We have considered sev-eral representative fuel cycles: light water reac-tors and more advanced thermal reactors andassociated fuel forms, operated in an open,once-through fuel cycle; closed cycle with Purecycling in the PUREX/MOX fuel cycle; andclosed fuel cycles based on fast reactors andactinide burning. The priority concern isaccounting and control of weapon-usable mate-rial during normal operation anddetection/prevention of process modificationor diversion to produce or acquire such materi-al.2

The open fuel cycles seek to avoid the prolifera-tion risk of separated plutonium by requiringthat the highly radioactive spent fuel beaccounted for until final disposition. Thisdefines the baseline for adequate proliferation-resistance, assuming that spent fuel is emplacedin a geological repository less than a century orso following irradiation (i.e., before the self-protection barrier is lowered excessively).However, the open fuel cycle typically requiresenriched uranium fuel, so the spread of enrich-ment technology remains a concern.

The advanced closed fuel cycles that keep theplutonium associated with some fission prod-ucts and/or minor actinides also avoid “directlyusable” weapons material in normal operation,since there is a chemical separation barrier anal-ogous to that which exists with spent fuel.Nevertheless, closed fuel cycles need strongprocess safeguards against misuse or diversion.However, the development and eventualdeployment of closed fuel cycles in non-nuclearweapons states is a particular risk both from theviewpoint of detecting misuse of fuel cycle facil-ities, and spreading practical know-how inactinide science and engineering.

Greater proliferation resistance will require theadoption of technical and institutional meas-ures appropriate to the scale and spread of theglobal growth scenario and responsive to bothnational and sub-national threats. Proliferationconcerns contributed significantly to our con-

MIT_ch08_65-70.qxd 7/16/2003 1:44 PM Page 67


clusion that the open, once-through fuel cyclebest meets the global growth scenario objec-tives, since no fissile material easily usable in anuclear weapon appears during normal opera-tion, and the “back end” does not have plutoni-um separation facilities. Enrichment facilitiesthat could be employed for HEU productionrepresent a risk. A variety of measures can min-imize the risk: strengthened IAEA technicalmeans to monitor material flows and assays atdeclared facilities; reliable supply of fresh fuel(and perhaps return of spent fuel) from a rela-tively small set of suppliers under appropriatesafeguards; implementation of IAEA preroga-tives with respect to undeclared facilities (the“Additional Protocol”); strengthened exportcontrols on enrichment technologies and asso-ciated dual-use technologies; and utilization ofnational intelligence means and appropriateinformation sharing with respect to clandestinefacility construction and operation. This is ademanding agenda, both diplomatically and inits resource needs, and calls for active effort onthe part of the U.S. and other leading nuclearcountries. With such an effort, the level of pro-liferation risk inherent in the possible expan-sion to 1000 GWe nuclear power by mid-centu-ry appears to us to be manageable.

It is clear that international RD&D on closedfuel cycles will continue and indeed grow overthe next years, with or without U.S. participa-tion. We believe that such work should berestricted by proliferation considerations tothose fuel cycles that do not produce “directuse” nuclear materials in their operation.Current R&D planning discussions in the U.S.reflect this concern. Such fuel cycles may alsohave manageable proliferation risks when cou-pled with improved technical and institutionalsafeguards. However, although advanced closedfuel cycles cannot realistically by deployed formany decades, the R&D program could itselfassist and provide cover for proliferants unlessstructured carefully from the beginning. Today,the international discussions are carried out bythose principally interested in developingadvanced technologies, without the neededlevel of engagement from those whose primary

responsibility is nonproliferation. The U.S.could play a crucial role in shaping these discus-sions properly before major efforts are under-way.

In this context, the PUREX/MOX fuel cycle is amajor issue. It is the current candidate, becauseof experience, for near-term deployment innations determined to pursue closed fuel cycles.However, it should be stressed that thePUREX/MOX fuel cycle is not on the “technol-ogy pathway” to the advanced fuel cycles dis-cussed earlier (typically, the advanced fuelcycles will involve different separations technol-ogy, fuel form, and reactor). The U.S. shouldwork with France, Britain, Russia, Japan, andothers to constrain more widespread deploy-ment of this fuel cycle, while recognizing thatdevelopment of more proliferation-resistantclosed fuel cycle technologies is widely viewedas a legitimate aspiration for the distant future.The associated institutional issues encompassexamination of the underlying internationalregime embedded in the NPT/Atoms for Peaceframework. All of these issues confront the fun-damental question of tradeoffs of national sov-ereignty in the context of access to nuclearmaterials and technology. Such issues areintrinsically difficult and time-consuming toresolve through diplomacy, but concomitantlyimportant for realizing the global growth sce-nario, while preserving international commit-ment to and confidence in a strong nuclearnonproliferation regime.

In summary, the global growth scenario built pri-marily upon the once-through thermal reactorfuel cycle would sustain an acceptable level of pro-liferation resistance if combined with strong safe-guards and security measures and timely imple-mentation of long term geological isolation. ThePUREX/MOX fuel cycle produces separatedplutonium and, given the absence of compellingreasons for its pursuit, should be strongly dis-couraged in the growth scenario on nonprolif-eration grounds. Advanced fuel cycles mayachieve a reasonable degree of proliferationresistance, but their development needs con-stant and careful evaluation so as to minimize

MIT_ch08_65-70.qxd 7/16/2003 1:44 PM Page 68


risk. The somewhat frayed nonproliferationregime will require serious reexamination andstrengthening to face the challenge of the glob-al growth scenario, recognizing that fuel cycle-associated proliferation would greatly reducethe attraction of expanded nuclear power as anoption for addressing global energy and envi-ronmental challenges.

NOTE

1. “DOE’s Nonproliferation Programs with Russia, HowardBaker and Lloyd Cutler, co-chairs, Secretary of EnergyAdvisory Board report, January 2001;“ControllingNuclear Warheads and Material”, M. Bunn, M. Wier, and J.Holdren, Nuclear Threat Initiative report, March 2003.

workshop materials,” Los Alamos Report — LA-CP-03-0233, (December, 2002).

MIT_ch08_65-70.qxd 7/16/2003 1:44 PM Page 69

2. E. Arthur, et. al.,“Uranium enrichment technologies:

MIT_ch08_65-70.qxd 7/16/2003 1:44 PM Page 70

71C h a p t e r 9 — P u b l i c A t t i t u d e s a n d P u b l i c U n d e r s t a n d i n g

There is little question that the public in theUnited States and elsewhere is skeptical ofnuclear power. A majority of Americans simul-taneously approve of the use of nuclear power,but oppose building additional nuclear powerplants to meet future energy needs. Since theaccident at the Three Mile Island power plant in1979, 60 percent of the American public hasopposed and 35 percent have supported con-struction of new nuclear power plants,although the intensity of public opposition haslessened in recent years.1 Large majoritiesstrongly oppose the location of a nuclear powerplant within 25 miles of their home.2 In manyEuropean countries, large majorities nowoppose the use of nuclear power. RecentEurobarometer surveys show that 40 percent ofEuropeans feel that their country should aban-don nuclear power because it poses unaccept-able risks, compared with 16 percent who feel itis “worthwhile to develop nuclear power.”3

Why does nuclear power, or for that matter anyenergy source, receive or lose public confi-dence? There is a surprising lack of survey datain the public domain that would allow us tounderstand why people oppose and supportspecific power sources. 4 To fill that void, wehave conducted a survey5 of 1350 adults in theUnited States. This internet survey6 measurespublic opinion about future use of energysources, including fossil fuels, nuclear power,hydroelectricity, and solar and wind power.

Our survey showed the same level of skepticismas other surveys. Respondents in our survey, onaverage, preferred that the United States reducesomewhat nuclear power usage in the future.The same, however, was true of coal, the

nation’s largest energy source, and oil. On aver-age, respondents wanted to keep natural gas atits current level. And, respondents strongly sup-port a significant expansion of wind and solarpower.

On what do these attitudes depend? Weexplored this question this question two ways.First, we performed a statistical analysis todetermine which factors explain who supportsnuclear power and who does not. This analysisis presented in the Chapter 9 Appendix. Theresults are, briefly, as follows:

Perceived environmental harms weigh mostheavily. The average person responded thatnuclear power is moderately harmful to theenvironment, and the difference betweensomeone who perceives nuclear power as“somewhat harmful” and “moderately harm-ful” is the difference between wanting toexpand and wanting to reduce nuclear powerin the future.

Safety and waste are also significant factors.Those who believe that waste can be storedsafely for many years express higher levels ofsupport for building additional nuclearpower plants. Those who believe that a seri-ous accident is unlikely in the next 10 yearsalso express higher support for nuclearpower. The problem is a majority of respon-dents do not believe that nuclear waste canbe stored safely for many years, and the typ-ical respondent believes that a serious reac-tor accident is somewhat likely in the next 10years.

Perceived costs of nuclear power are thethird most important factor. Those who

Chapter 9 — Public Attitudes and Public Understanding

MIT_ch09_71-74.qxd 7/16/2003 1:47 PM Page 71


power and somewhat more support for coaland oil. Information about global warmingagain had no effect on public attitudes towardalternative energy sources.

In our view, these survey data reveal the funda-mental importance of the technology itself forpublic support. American public opiniontoward energy is not the product of politicalideology or party politics. Rather, public oppo-sition to nuclear power in the United States isdue primarily to the public reaction to the con-crete problems of the technology and theindustry, notably concerns over safety, toxicwaste, and poor economics. It is not surprisingthat the public is skeptical about a technologythat has over promised.

Should there be a public campaign to changeperceptions about nuclear power? The evidencesuggests that such a campaign may have onlymodest effect. Most of the change would comethrough education about the high price ofalternative energy sources, such as solar andwind. The other possible source of change inpublic attitudes is the connection betweenglobal warming and fossil fuels. The typicalperson expresses concern about global warm-ing, but that concern does not in turn translateinto higher support for carbon free electricitysources, such as nuclear power.

The surer way to cultivate public acceptance ofnuclear power, though, is through the improve-ment of the technology itself and choosingcarefully what nuclear technology to use.Developing and deploying technology thatproves uneconomical and hazardous will makethe global growth scenario infeasible.Technology choices and improvements thatlower the cost of nuclear power, that improvewaste management and safety, and that lessenany environmental impact will substantiallyincrease support for this power source.

believe nuclear power is uneconomical sup-port it less.

Surprisingly, concern about global warming,in our survey, does not predict preferencesabout future use of nuclear power. There isno difference in support for expandingnuclear power between those who are veryconcerned about global warming and thosewho are not.

Political beliefs and demographics, such asage, gender, and income, mattered relativelylittle, if at all.

Second, we performed an experiment withinthe survey to measure sensitivity of attitudes topossible changes in cost, waste, and globalwarming. Half of the sample was provided noinformation; they are the control group. Theremaining half was divided into four groups.These groups were provided with informationabout future energy prices or about toxic wastefrom fossil fuels or about global warming orabout all three factors (economics, pollution,and global warming). Our aim was not toincrease support for nuclear power, but to seehow the mix of energy sources would changewith accurate information about costs, toxicwaste, and global warming.

Only nuclear power showed substantially moresupport between the control group and the oth-ers. Those who received all three pieces of infor-mation supported nuclear power and naturalgas equally, and supported nuclear power muchmore than coal and oil.

Information about the relative prices of energysources produced almost all of this shift. Thepublic perceives solar and wind to be inexpen-sive. When informed that solar and wind aremore expensive than fossil fuels or nuclearpower, survey respondents showed substantial-ly less support for expanding solar and windand substantially more support for nuclear

MIT_ch09_71-74.qxd 7/16/2003 1:47 PM Page 72

73C h a p t e r 9 — P u b l i c A t t i t u d e s a n d P u b l i c U n d e r s t a n d i n g

1. Eugene A. Rosa & Riley E. Dunlap,“Poll Trends: Nuclearpower — three decades of public opinion” PublicOpinion Quarterly, 58, 295-324 (1994). National ScienceBoard, Science and Engineering Indicators 2000, volume1, page 8–19. Washington DC: National ScienceFoundation. Survey results vary because researchers askdifferent questions and in different contexts. Recent sur-veys range from 60 percent opposed to “building newnuclear power plants” (AP/Washington Post) to 55 per-cent favoring “new nuclear power plants in the future”(Nuclear Energy Industry tracking survey questionnaire,October 2002, carried out by Bisconti Research Inc). Fora discussion of some of these issues and the state ofpublic opinion, see Steve Miller “Pragmatic ConcernsFuel Nuclear Support,” IEEE Spectrum, http://www.spec-trum.ieee.org/WEBONLY/publicfeature/nov01/natt.html.Because existing survey data do not directly addressmany of the issues that motivate our inquiry, we con-ducted our own survey.

2. Associated Press poll conducted by ICR March 12–161999, N = 1015 adults nationwide. MIT Energy Survey,June, 2002, N = 1350.

3. European and Energy Matters, 1997, EUROBAROMETER46.0, Directorate General for Energy, EuropeanCommission, February 1997.

4. Studies of particular factors have been conducted.Accidents and waste loom large in public thinking. See,for example, Ellen Peters and Paul Slovic, Journal ofApplied Social Psychology, 26, 1427-1453, (1996). Conniede Boer and Ineke Catsburg,“A Report: The Impact ofNuclear Accidents on Attitudes Toward Nuclear Energy,”Public Opinion Quarterly, 52, 254-261 (1988).

5. We surveyed the United States for reasons of cost. A reli-able survey of a similar size in another country per-formed by a reputable survey research firm was tooexpensive. It is our hope that this survey offers a model

for studies of public attitudes toward energy use anddevelopment in other countries. The responses might bequite different. For example, Europeans are more con-cerned with global warming which could influence theirattitudes toward nuclear energy.

6. We performed an Internet based survey because of fourdesign advantages over the alternative methods, phoneor face-to-face. First, a face-to-face survey was prohibi-tively costly — at least 10 times the cost of the Internetsurvey. Second, Internet surveys have much higherresponse rates than phone surveys. KnowledgeNetworks, the firm we employed, recruits a pool ofapproximately 2 million people from which it draws arandom sample. Approximately 80 percent of the peoplesampled responded to our survey within one week. Thetypical phone survey with a similar cost structure has anon-response rate of around 70 percent. Third, ensuringa higher response rate in a phone survey would haveincreased costs substantially (approximately double).Fourth, Internet surveys are ideal for the experimentalmanipulations we performed. We provided informationin graphics and text format, which is superior to readingtext over the phone.

The drawback of the Internet survey is that Internetusers are not necessarily representative of the popula-tion. Knowledge Networks recruits a pool of potentialsurvey respondents from the general population anddevelops sample weights to allow us to extrapolate tothe general population. So, a college educated, highincome individual receives less weight than an individ-ual without a bachelor’s degree and with modest or lowincome, because individuals with college educationsand above average income are more common in thepool than in the population. Data analyses are per-formed with appropriate sample weights and control-ling for demographic factors.

NOTES

MIT_ch09_71-74.qxd 7/16/2003 1:47 PM Page 73

MIT_ch09_71-74.qxd 7/16/2003 1:47 PM Page 74

Documents

Chapter 4 â€” Fuel Cycles - MIT - Massachusetts Institute of