. ORNL-2985
~ UC-81 - Reactors-Power

 

S _THE FEASIBILITY OF AN UNATTENDED
NUCLEAR POWER PLANT
-‘,':‘:_'M W. Rosenthal
"M.. M. Yarosh

C, F. Baes, Jr. |
~ R.C. Robertson -

et i A LT p e e e e b e

 

 

 

 

| |
OAK RIDGE NATIONAI. I.ABORA‘I'ORY

_ | opercted by , '

“UNION. CARBIDE CORPORATION

e for the -

u 5. A‘I’OM!C ENERGY commssuou o

|

b

 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

$1.75
Printed in USA. Price L e Avellable lrom the _

thce ol Technical Servlces '
. \Depenment of Commerce . _
o Weeblng!on 25, D.C. .

 

 

 

 

7 Tl’us reporf was’ prepared as an occounl of Governmenf sponsored work. Neuther vl-e Um!ed Sfo!es,.' -

AL Mckes any warronfy or represeniahon, expressed or |mpl|ed with respect ‘to the uccurucy,-_

" B.:Assumes any liabilities: wnh respect to the use’ of or fof domuges resulflng from the use ol o

.,As used: in the nbo\ve, ‘person nctmg on behalf of the Commission®” includes any employee or -

'contracfor of the Commsslen, or employee of such contractor, to the extent that such employee
or confrac!or of ‘the Commus:on, or employee of luch contractor - prepares, d:ssemmates, or

Hfl'provldes access to, ‘any “information pursuonl to l'us employmem or t:onirm:f with 'he Commus:on,

. or his employment wnh such contrccfdr. STl e

 

~LEGAL NOTICE = -

nor ?l'ne Cemmlsslon, nor any person. achng on behulf of the Commissiont
o 'completeness, or usefulness _of ‘the mformaflon confmned in fhls raport,. or |l'|ut the use of
“any information,’ opparutus, meihod .or precess d;sclosed in tlns repon may nof mfrmger'.

_‘prwately owned rights; or °

- any. mlormahon, apporatus, methed, or pfocess dlsclosed in this reperI. B

 

 

 

 

 

 

 

 

 

 

 

 
 

0

ORNL~-2985
Reactors~Power
TID-4500 (15th ed.)

Contract No. W-ThO5-eng-26

THE FEASIBILITY OF AN UNATTENDED NUCLEAR POWER PLANT
M. W. Rosenthsal
M. M. Yarosh

C. F. Baes, Jr.
R. C. Robertson

DATE ISSUED

SEP§ 1960

 

OAK RIDGE NATTONAL LABORATORY
- _Oak Ridge, Tennessee
- operated by
UNION CARBIDE CORPORATION .
- .. for the
U.S. ATOMIC ENERGY COMMISSION

 

 

 
 

TS
 

 

 

a*

&

TABLE OF CONTENTS

A‘bstract ..."........O‘."'....'.,.....C.......-..."'..t....l...l

INTRODUCTION ..ceeaccscacoosasssssssnsoncnssanesssconscasosoosass
SOME COMMENTS ON THE ACHIEVEMENT OF RELIABILITY .veevvosocscossnse
SELECTION OF A REACTOR POWER PLANT ..... creesne cesesssessenssannn
Reactor Type ...................................;................
Power Recovery System .ceeseeecescescssscssessscscsescasscscssssnns
CONCEPT OF A SIMPLIFIED PRESSURIZED-WATER REACTOR ..cicecaccccacs
Primary System cieecereecreaesssecsessscssssssscanansessssasosssens

Reactor Control ..eacsvcecsecosnnsacans cresesstacsrrenroeann
Fuel Elements ..ciiesessssscasacosvsccssssossrscoasascsssssss
Primary Coclant PumlpPS .eueesvevescssessscnssssscsassssscoasnss
Core PressurizZel viseeececssrsecnssscsscsncesssssnssssnscsnnes
Main Heat EXChangeYr .ceveesessscsecasccscosesssasessssscasans
Water Treatment ..ccisceccesancrsssscescesscscnascanscscnssss

Second—ary System & 9SO H b e EF RS eE e hd kSt bR S A eSS st

Simplified Conventional System .....cceveescesecasncesassasesns
Semiconventional Steam System ..ieeieeeecscecnsssacssoncsans
Hermetically Sealed Steam System .veececesccccssssstcsssocss
Basic Features of the Steam CYCle .vvveesecsesssassssoccsass

EFFECT OF ALTERED REQUIREMENTS ON THE DESIGN CONCEPT ...cceeeeeee
Communication with the Plant ....ceceerersconsssacssosscascnsnses
Plant Size and Welght ...veeevaeracscsescnccscarscsnsccsncsssssne
Long~Term Rellability and Expendabllity ....cceeravearerescsencnn
VmflflfleLmfls.“.“.n.“.“.n.“.".“.".u.u.g.n.“.“..
Effects of Time Schedule ..c.iicecescesscoveasssssssssanssansnnens
SUMMARY AND comcwsxons
REFERENCES sesesesernsesesncatasrersaranscicrisuortsesatesrevanons

13
13
15
16
17

17
24
26
27
30
30

33

34
35
36
3%

53
53
54
55
55
56
56
60

iii

 
 

 

- -

i

THE FEASIBILITY OF AN UNATTENDED NUCLEAR PCWER PLANT

Abstract

A study has been made of the feasibility of constructing
a small nuclear power plant capable of operating one year
completely unattended. A system which will produce 1 Mw of
electricity without interruption is required for field use in
about four years.

It is the opinion of the authors that the four~year
requirement greatly restricts the opportunity for develop~
ment of new concepts and new technology and that the objective
is most likely to be attained by extreme simplification of a
type of reactor with which there has been favorable experience.
A pressurized~-water reactor was selected for the application,
both because of the extensive experience with it and because
it appears readily adaptable to simplification. For example,
the power plant suggested has hermetically=~sealed primary
and secondary systems, and all control systems, except the
turbine governor, have been eliminated.

The guthors conclude that it is feasible to develop a
relisble, simplified, pressurized-water reactor system for
unattended service in the allotted time.

INTRODUCTION

A study was undertaken by staff members of the Osk Ridge National
Laboratory to evaulate the feasibllity of constructing a small nuclear
power plant capable of operating unattended for ome year. The objectives

of the study were, specifically, (1) to assess the potential for success

of a program'to design.and develop such a- system.within & limited time
w;iperiod and (2} to propose, if possible, concepts which offer promise of
, fmeeting the design objectives. The 51ngulsrly'most important consideration
;,[fwas the- degree of unattended reliability achlevable from the indiv1dual

'“~igffigcomponents of a nuclear plant and fromfthe combined operating SYStem~

In estlmatlng the potentlal for reliability of a nuclear power plant,

_,_:it was necessary‘to conceive of a design capable of saxisfylng the system
_,;reqpiremsnts.L The primary objective, however, Was estimation of the |
S potential for success of a program based on such a design, rather than

. presentation of the de51gn itself.

 
 

 

 

"(/.‘

- As will be evident throughout the report, the design consideratlons
have been restricted to meeting only the stated objectives of the.study. (;;
It was the opinion of the authors that the achievement of one year of
unattended operation was a problem of such difficulty as to dominate all
‘other design considerations. Thus the question, what makes a plant
cease to function, was continually in the forefront throughout the study.
That this apparent obsession with reliability is a reqpisite may be |
atteéted to by the experience records of those reactors which_are'operating
today. - | _ | , o a
For the present study'the following system requirements wefe,_
established: | B
1. The plant is to operate.unattended at full power continuously
for a minimum of one year. ' | |
2. The net electrical output is to be 1000 kw of three-phase alter-
nating current at 4160 v.
3. The frequency can be selected by the designer within a range
of 60 to 1000 cycles, but the frequency selected must be held to within
+1% throughout the life of the plant. |
4. The reactor plant will be the only power source connected to
its electrical load.
5. A reactor power plant capable of satisfying the preceding
requirements is to be in operation on iocation within approximately
four years.
6. Interruption of the generation of electricity for any reason =
within one year of initiation of operation is considered a plant failure.
7. Short=term (one=year) reliability is of utmost importance and | .
will not be sacrificed to provide long~term life.
8. The system could be considered expendable, if necessary, at
‘the end of one year of operation. .
_ 9. During startup and shakedown of equipment, semiremote operation
of the plafit will be possible, and special equipment may be provided for

startup requirements.
 

Fil

a

10. After the beginning of unattended operation, no commmnication
with the system willrbe possible other than knowledge that the electrical
load is being supplied.

11. Within.reason, plant size and weight are not considerations.

12. Within reason, cost is not a consideration.

13. Plant efficiency is not an important consideration.

14. The system will operate stationary.

15. The plant will not operate in a populous area.

16. The sink for heat rejection will be determined by the appli=-
cation, but adequate means of cooling will be avallable.

The time limitations lmposed on this study precluded an extensive
examination of all systems which.night conceivably satisfy the require=
ments., Hence, only those systems that appeared to offer the maximum
potential for reliability were studied. There is no intent to imply
that the systems discussed in this report comprise the only ones capable
of satisfying the plant requirements, but, within the framework of
religbility considerations for this system, they appear to be the most
promising. |

The authors are deeply indebted to various members of the Oak Ridge
National Laboratory and the Osk Ridge Gaseous Diffusion Plant staffs for
numerous contributions to this study. L. D. Schaffer helped apprise the
authors of the status of smsli-reactor development. E. R. Mann advised

the authors on'reactor control' Ww. C. Thurber and P. Patriarcs were

_consulted on, metallurglcal problems and the selection of fuel elements.
- E. J. Breeding and W. G Cdbb were consulted regardlng bearlngs and
,;seals, J.AL Gabbard regarding electrical generators, and P. H, Pitkenen
-d‘dregarding core physics';; e '

Helpful diSCussions were held W1th W; B. Cottrell, J. E. Cunnlngham,

AP Frass, C. i—'fH. Gabbard, E. Be Gross, C. J. _Hochenadel, P. G. Lafyatis,
. H.C. McCurdy, H. F. McDuffie, A: M. Perry, I. Spiewsk, E. Vincens, C. S.
”"ffWalker, and members of the Englneering Development Department Technical

 

ifi?f;:;D1V151on, Oak Rldge Gaseous Diffu51on Plant Burns and Roe, Inc., con-
'isultlng englneers to the L&boratory, studled certaln problems of the

 
 

 

 

secondary system and contributed information pertinent to component
selection and reliability.

During the course of the study, discussions and correspondence were
carried on with personnel of a number of organizations. Many.of the
opinions expressed in this report are a result of such discussions, and
+the authors gratefully acknowledge the following organizations for their
valuable contributions:

Aerojet General Corporation (GCRE)

Alco Products (Dunkirk Facility)

Allis~Chalmers Manufacturing Company

Atomics International

Combustion Engineering (Windsor Facility and SL~1)

The Elliott Company (Jeannette, Pennsylvania)

General Electric Company (Erie, Pennsylvania; Fitchburg, Massa~-
chusetts; and Schenectady, New York)

Gilbert Associates

International Nickel Company

Maxrtin Company (Baltimore, Maryland)

Personnel of SM=1 and GITF, Fort Belvoir, Virginia
Pierce Governor Company

Westinghouse Electric Corporation (Bettis Plant, East Pittsburgh,
and Lester, Pennsylvania.)

Woodward Governor Company
Worthington Corporation (Harrison, New Jersey)
Gilbert Associates kindly made available to the authors a draft of their

forthcoming report1 on the reliability of reactor components.

SOME COMMENTS ON THE ACHIEVEMENT OF RELIABILITY

In considering the objectives of this study, it is meaningless to
speak of building a reactor system which will operate without failure
for one year. To do so would suggest we are requiring it to be a cerw
tainty that a particular reactor power plant operate successfully for

that period. Actually, a more realistic specification might be that we

LH

-

 
 

 

)

i

»

s

expect nine out of ten reactors to be operating at the end of a year,
which is equivalent to requiring that an individual reactor have a
reliability of 0.9. For purposes of this study, it was assumed that a
plant reliability of 0.9 was required for one year of unattended operation.
Some understanding of what is involved in achieving a 0.9 reliability
may be obtained by considering how the reliability of the individual com~
ponents affects the over=all reliability of the power plant. If a system
consisting of a number of components is to have a reliability of 0.9,
the reliability of each component in the system must be better than 0.9
and an average component reliability very much better than 0.9 may be

necessary. This is illustrated by Eq. (1) in which p; 1s the probability
that component "i" will last one year, and P is the probabllity that all
the components will last one year:

P=P1PpD3 « « « D - (1)

For example, if the system consists of 10 components, each having a
reliability of 0.99, the probability of the system lasting one year
without failure of any of the components will barely exceed 0.90. If
there are 50 components, the failure of any one of which will stop the
functioning of the system, the "average" reliability of each component
must be 0.998. The significance of a 0.99 requirement may be more
clearly understood if one pictures & test system in which 100 identical
components are simultaneously set into operation. For a 0.99 relisbility,
99 of the 100 components must still be operating at the end of a year.
Actually; the prdblem is more complicated than Just assessing the
probability that a particular,reactor can operate unattended for one
year. No-nuclear power plant having"the needed capacity has been built
to operate without regular maintenance. For reactors such as the M-l
(APPR) and the SL-1 (ALPR), it is en accomplishment to operate 1000 hr
without shutdown, even with maintenance operations being performed during
that period. Hence, it is. not a question of whether an eXisting system
can meet the spe01ficat10ns, but whether a modification of an existing
system or a new design of reactor can‘be made suffiCiently reliasble for

unattended operation. The problem is to estimate the likelihood that a

 
 

 

 

 

 

reactor system having the necessary reliability can be developed in
about four years.

' In this study we have attempted to distinguish between (1) the
probability that the development problems associated with unproven con=
‘ceptsror untested equipment‘can-be'solved in a limited time period, and

(2) the probability that a particular system based on tested equipment
.and proven concepts can operate for a year unattended. For convenience,
we shall designate the first probability as P; and the second as P,.

| One can conceive of systems which, because of inherent features,
appear to be capable of long periéds of unattended operation (have a
high value of P,) after some important'development problems assoclated
with them are solved. An example might be a natural=-convection liquid-
fuel reactor‘that uses thermoelectric devices for the generation of
electricity. There would be no moving parts to wear and no control
syétem to fail. A power plant of this type, if developed successfully,
would probably operate unattended for long periods. However, there
.~ would be an appreciable risk (low Pl) in proceeding to develop it for
attainment of the objectives, since some of the development problems
might still be unsolved at the end of four years.

On the other hand, a reactor very similar to the SM~Ll reactor
might be used. If no major changes were made in the control system,
water~treatment system, etc., one could be confident that a reactor of
this type could be bullt within the time allowed and that it would
operate at its design capacity, but it would not operate very long with=-
out attention. This reactor concept might be said to have a high P;
value but a low P, value. Since it is the product of Py and P, that
gives the pfdbability of successfully developing a system of the

required reliability in the time allotted, close attention must be given

to each in selecting a program to follow,

The‘language of mathematical probability may help to clarify the
problem, but the difficulty of evaluating the probabilities remains.
The unéertainty in P, is illustrated by several reactor development

progrsms which have not attained their objectives after a number of

*)

 
 

N

At

)

years of effort. 'While solutions may eventually be found to their
problems, a requirement of success in a limited time would not have been
met. Estimation of P, i1s also uncertain, since it requires knowledge

of the reliability of a number of components. Although the reliability
of a component which has been extensively tested may be known with some
assurance, even a small change in design can alter the value importantly.

Thus little confidence would be placed in an estimate of the reliability

- of a complex piece of equipment not yet designed.

In order to assess the reliability of existing equipment, operation
reports on a few reactors (SM~l, SL-1, PWR) were perused in order to
determine the frequency and causes of failures, and operating experience
was discussed with personnel at several installations (SM-l, SL-l, GCRE,
GITF). A study was also made of the operation history of a number of
moderate~sized steam turblnes.

The reactors examined were not meant to operate unattended, and
they were designed for regular maintenance. Hence, it was expected that
these systems, having hundreds of mechanical components and large numbers
of electronic devices, would be far less reliable than required for one
year of unattended service. More surprising and sobering was the con=
clusion that few existing components in normasl use would operate for one
year without maintenance., This conclusion is least applicable to the
mechanical equipment in the primary system of pressurized-water reactors,
because intensive effort in the Navy program has produced reliable com-

ponents. It applies very strongly, however, to steam power equipment,

~ where spares ere'provided for'many 1tems, and appreciable on-stream

_maintenance is the normal practlce.a Electronic control devices that use .

vacuum tubes are also quite unreliable for one year of service. An

extreme 1llustrat10n is prOV1ded by the SM~1 control system, in which

»300 vacuum.tube replacements were made during the first year of operatlon.

The data on. turbine-generators were of particular interest because

'_they 1ndicate how rellabillty can be 1mproved From a study of the

lA-year operating history of a. number of conventlonal 25-Mw turbine~

'generator_units,:Myers concluded that - only five units in 100 would be

2

 
 

 

capablé of operating continuously for one year. Myers' study was
extended to three 3000~kw turbine-generator units, and a value of 0.05

for the reliability was again obtained. Further examination of the

failures of the smaller systems indicated, however, that if the need for

replacing the generator brushes and repacking the admission valves could

be eliminated, approximately 60 units in 100 could operate continuously

- for one year.

There are three ways one might proceed to improve the reliability
of a system which is to operate-finattended: (1) eliminate unrelidble
components by eliminating their functions, (2) improve the reliability
of the equipment by design and development, and (3) duplicate less
reliable items so that if they fail, spares will automatically continue
to perform the function.

- Many components in reactor power plants appear to perform functions
that ‘are not necessary under the ground rules of this study. An example
is an overload protection device, which will sometimes interrupt service
unnecessarily and hence 1s an undesirable item with regard to achievement
of uninterrupted operation. Where continuous power production is the
only eriterion by which performance is Jjudged, stoppage of a pump because
the circult breaker trips is a plant failure equivalent to stoppage
because the windings burn out.

The reliability of components can be improved in several ways.
Modifications, such as substitution of better materials, use of better
methods of fabrication, tighter quality control and inspection, and more
extensive testing of an item before acceptance, may increase the probable
life of an existing device. In addition, the change from normal require=-

ments resulting from the specific objectives of the design may permit the

‘use of a different type of device to perform the required function. As

pointed out in the Gilbert Associates study! of reactor reliability, the

achievement of reliability is not consistent with a minimum~cost philosophy.
In considering the use of spares, an important point to recognize

is that improved reliability does not necessarily accompany duplication.

An insufficiently reliable control device for switching to a duplicate

.

 
 

 

o

»

ju

»

component, for example, could reduce the over;all reliability of the
system. Duplication of a component which might develop & leak doubles
the chances of that type of failure.

One way to improve reliability, particularly in control equipment,
is to use coincidence circultry which requires, say, that two out of
three parallel devices operate before the function is performed. With
this type of circuitry, spuriocus operation of one device does not cause
the function to be performed, and fallure of one device to operate does
not prevent performance of the function. It 1s worth noting that a
system.of this type works best under supervision so that a component
which has failed can be repaired to return the system to its initial
capability. In principle, however, the system can be made as elaborate
as required to achieve the needed reliability.

It is likely that reliability requirements on individual components
can be relaxed somewhat if advantage is taken of the concept of operating
spares. For example, suppose that a system has w0 circulating pumps,
both operating all the time, but each having a capacity such that one
alone will provide sufficient flow for reactor operation. With this
system the reactor is not prevented from continuing operation by the
failure of one pump. If each of these pumps has a reliability of 0.9,
then the probabllity, as computed from the following equation, is 0.99
that at least one of the pair will still be operating at the end of a

year:

Pg=l-Q-»", | (2)

where P is the probdbility of survival of at least one component in the
group, p is the probabllity of survival of an individual component and
nis the number of identical components. If there are three pumps, each
having a 0 9 reliability, then the prdbdbility is 0.999 that at least one
will be operating at the end of & year.‘

| The time allowed for the development of a reactor system.has 8 very
strong effect on the selectlon of a program.f As stated in the Introduction,

this study presumes there ig need‘for & reliable reactor to be in the

 
 

 

 

 

field in about four years. If s prototype reactor is to be operated
before the first field installation is made, and this seems essential,
four years is a fairly short time for the program. The time required
. for construction of a small reactor, excluding development time, is
indicated by the experience with three reactors in the 1-Mw (electrical)
range! |

1. The SM~l (APFR, pressurized-water reactor built by Aleo Products
at Fort Belvoir) was eritical 29 months after award of the contract. A
conceptual design, which included & fairly detailed description of the
'core, was available before the contract was awarded.

2. The PM~24 (pressurized=water reactor for polar region, also
bullt by Aleco Products) was assembled in the manufascturer's plant 14
months after award of the contract. A fair amount of design work had
been done before the contract date. This reactor is & skid-mounted airw
transrortable adaptation of SM=l. ’

3. The PM~1 (pressurized-water reactor being built by Martin Company
for location at Sundance, Wyoming), as presently scheduled, will require
about 27 months from the beginning of a parametric study to assembly‘of
the reactor. About six months of this was required for final design,
and about 15 months is estimated for the period from heginning of order
placement to assembly of the system. The PM»l is an advance from the
EM~l, and it uses & new fuel element design. TFuel element development
and physics studies began considerably in advance of the 27-month period
referred to above.

In the last two cases, it should be noted that the times given are
for assembly of the reactor. They do not allow for testing before going
to power. _

’ If it is presumed that a program is to be undertaken with some
urgency and a falrly liberal budget, based on the figures for the three
reactors above and onvdiscussions with reactor manufacturers, the timé
_réquired for design and construction of a prototype reactor (built for
high reliability) appesrs to be about two years, once the design concept
'is established. Allowing six months for finishing the conceptusl design

10

N
 

 

 

 

wh

O

A

and making arrangements for the detail design and manufacture, the
prototype might be ready for testing in about two and one~half years,
as shown by Fig. 1.

UNCLASSIFIED
ORNL-LR-DWG 501064

YEARS
o | 2 3 4

 

CONTRACTOR
SELECTION
)
CONCEPTUAL DETAIL DESIGN
DESIGN J AND CONSTRUCTION OF PROTOTYPE OPERATION OF PROTOTYPE

 

J
A

 

 

CONSTRUCTION OF
FIELD REACTOR

 

 

 

 

 

 

 

ORDER LONG LEAD ITEMS FOR FIELD REACTOR

DECISION TO INITIATE PROGRAM
BEGIN DESIGN OF FIELD REACTOR

Fig. 1. Time Schedule for Achievement of Field Reactor in Four Years.

The design of the first reactor for field use might begin before
completion of the prototype, and all orders for equipment for it placed
after one=half year of test operation. Allowance of one year for con~
struction of the field reactor would bring its completion time to four
years from the beginning of the program
' This schedule has no prOV1sion‘for developmental work other than
that which proceeds concurrently_with the design and construction of

'the'prototype; There'is no-contingency allowance for difficulties in
,fthe development of the prototype, nor is time permitted for the ‘solution
of the problems found during operation of the prototype reactor before

_'construction of the second reactor commences., Hence, even with very

little'de#elopment'rerired'snd'with good luck at every step, about four

| years would be reqpired to get the first reactor in the field. If there

were delays in administrative decisions, construction holdups, or

 
 

 

 

unforeseen development difficulties, well over four years might expire
before a reliasble power plant were achieved.

In view of the above, a development program based on the following
general principles appears to have the greatest promise of success in
producing a reactor to satisfy the reliebility requirements in a period
of about four yesars: ,

1. The main effort should he devoted to perfection of a system
based on concepts with which there has been favorable experience. This
restriction refers not only to reactor and power plant types in the broad
sense, but also to the manner in which the systems sre to be operated.

2. Approaches based on unproven concepts should be studied as
secondary progrems if they offer promise of giving a better system than
that based on proven concepts.,

3. Exceptions to points 1 and 2 could be made for parts of the
system 1if a switch in midprogram to & proven concept would not delay
the project. ‘ |

4, The most promising a@proach to the achievement of high relia-
bility is simplification. The objectives of this study make simplifi-
cation of the reactor particularly promising, since the system need not
be repairable, have & long life, follow a varying load, or operate in a
~populous area.

5. A close interdependence of the system parts is permissible,
since in any case a failure would terminate the ability of the system
to produce electricity.

6. Components should be overdesigned and underworked to increase
their life-expectancy. Efficiency should be sacrificed to gain
relisbility.

7. Components which are relatively unreliable can be used during
startup if they cannot affect the system once 1t is in operation.

8. The individual components and the assembled system should be
tested thoroughly before the system is left unattended. This testing
shouLd continue long enough to pass the period where early failures

resulting from manufacturing defects are likely to occur.

12

O

e

 
 

 

)

)\

The preceding principles can be summerized by saying that in seeking
the achievement of reliability in.the limited time available, one should
attempt to use proven concepts and present technology. The success of a
program to develop a reliable reactor in four years should not depend on
ma jor advancements in technology or the perfection of new concepts. If
more time were available, the restriction on new development would be
relaxed, but it would be stated even more strongly for a shorter time
allotment.

The approach described in this report may not lead to the best
system for unattended operation, Nevertheless it appears to be the
approach which has the highest likelihood of producing s reliable reactor

in four years.
SELECTION OF A REACTOR POWER PLANT

Reactor TXEe

Early in this study a comparison was made of a number of reactor
concepts with regard to their applicability to an unattended system.
These included the various liquid~fuel, organic~-moderated, gas~cooled,
and liquid~-metal~cooled reactors, as well as the pressurlized-water and
boiling-water variations of light~water-moderated reactors. Based on
the premises of the preceding section, the conclusion was reached that
the extensive and generally favorable experience with water~cooled
reactors recommends them for this application. . The potential advantages
of other systems, such as hlgh thermal effic1ency, favorable fuel or
neutron economy, low constructlon costs, and poss1bly light welght are
of secondary importance in the framework of this study.

o Once hav1ng selected a weter-moderated reactor as the basis for this

{ystudy, a further decision between ‘boiling-water and pressurlzed-water
‘a-systems was reqnired Although these reactor types have many features
'ein common, each has certaln advantages and dlsadvantages relative to the

| other The;obvlous advantage of the b01lingfweter system is the elimi-

netion of,eeveral major:itemS'of'eqnipnenti,che primary heat exchanger,

13

 
 

the main coolant circulgting pump, and the core pressurizer. An advantage

of the pressurized-water reactor is that there is no production of radio=- .

lytic gas. In contrast, the boiling~water reactors in operation at
present.produce radiolyt;c gas which is either vented to the atmosphere
~or recombined in a catalytic recombiner. Another advantage of the |
pressurized~water reactor is that reactor control'is accomplished very
simply, and there are no problems of stability. In addition, there has
been much more experience with pressurized~water than with boiling-water
systems, o ',

The heat exchanger on a pressurized-water reactor, although large |
and expensive, can be made quite reliable if care is taken in the selection
of materials and in fabrication. Pressurizers are not of themselves
basiéally unreliable, but the pressure control systems sometimes are,
1t appears to be possible to design a system in which the control, as
normally performed, is eliminated, thus meking the pressurizer function
reliable. The most questioned component is probably the primary cooclant
pump. Experience with canned=rotor pumps of the type needed indilcates |
that an individual pump that is carefully tested and found to be sound
will operate reliasbly well in excess of one year. In addition, the
concept of operating spares is directly applicable to the primary coolant
pump.

When there is an overpressure of hydrogen, radiolytically dissociated
water in a pressurized-water system is recombined internally with no
production of hydrogen and oxygen. All pressurized~water reactors are
operated in this manner. Although, in principle, water decomposition in
& boiling~water reactor can be suppressed by increasing the pressure,
operating at a high pH, and injecting hyrirogen,4 this mode of opereation
has not appeared attractive, and all bolling~water reactors in operation
evolve radiolytic gas. Besides the probiem of handling the radiolytic
‘gas,'the oxygen in the steam makes the materials problems more sevére in
' the'boilifig;water reactor than in the secondary system of a pressurized=-
water power plant.
| Highly enriched pressurized-water reactors can operate, and often

do (the SM~1 always), without automatic reactivity regulation, except

14

€

w

 
 

-}

>

for the inherent action of the negative temperature coefficient. As
discussed later, it appears to be feasible to operate a pressurized-
water system for one year with no movement of absorbing materials in the
core. This can be done using only proven concepts. Boiling~water reactors
normally are operated with continuous reactivity regulation by movement
of absorbing materials. Long~term operation without the use of an auto~
matic control system would depend on more of an innovation than in the
case of a pressurized-water reactor. (Boiling=water reactor control
might be considerably simplified, however, at steady power.)

Because it appears possible to make a simple pressurized-water
reactor with few innovations, the pressurized~water concept was made the
basis for this study. This selection was not made because the pressurized-
water reactor is inherently more reliable than the boiling-~water reactor,
but because there is more reason for confidence that a reliable, unattended,
pressurized~water reactor can be developed in four years.- There has been
extensive experience with pressurized-water reactors, and the method of
operation proposed here deviates little from the manner in which they

have been operated in the past.

Power Recovery System

 

The selection of a power recovery system to operate in conjunction
with the reactor plant was heavily influenced by the principles outlined
in the discussion on reliability. During the initial phases of this
study, consideration was given to the possibility of converting heat
energy to electrical power throfigh other than the use of a conventional
turbine-generator system. In particular, a brief study was made of

thermoelectric generators.; The conclu51on of the study was that, in

- view of the present early stage of development of thermoelectric materials

and the relatively restricted amount of operatlng experience, thermo~
electric generation does not appear to be promising for use at this time.>
The development of 1mproved materlals permltting higher efficiencles at
moderate tenperatures may in the future make such systems attractive for

unattended power stations, -

15

 
 

With the selection of a pressurized-water reactor and application
of the stated reliability concepts, it is clear that the most sultable
power system is one employing a conventional steam turbine~generator
set using water as the working fluid for the heat~power cycle. It is
appreciated that many of the corrosion problems associated with water
systems might be avoided with other flulds, but the technology and ex-
periencé with equipment using less corrosive fluids are not adéqpate for

a reliable design.

CONCEPT OF A SIMPLIFIED PRESSURIZED-WATER REACTOR

FOllowing the selection of the pressurized-water concept a critical
examinatidn of pressurized~water reactors was required to determine the
modifications necessary to achieve the required reliability. The advan=
tages of using proven concepts and present technology were balanced
against the galn to be made by a change in design and the likelihood of
the change being successful. Some components and procedures used on SMél,
for example, appear to be directly applicable to an unattended reactor.
In other cases, modifications are needed, and in a few instances the
development of new components would be desirable., Wherever innovations
are suggested in this report, there are alternate solutlons which could
be used should the new development not appear to be progressing rapidly
enough. All the new components proposed represent combinations of items
on which there has been successful experience,.

In the subsections which follow, the features of a pressurized-water
reactor that affect its reliability are discussed. This section does not
present a reactor design, but it does suggest concepts and techniques on

which & design could be based. A schematic diagram of a simplified

pressurized-water reactor is presented in Fig. 2 to indicate the type of

system which emerges from the discussion. All the components that function

during unattended plant operation are included.

16
 

 

 

Y

UNCLASSIFIED
ORNL-LR—DWG 50107A

PRESSURIZER ADMISSION

STEAM
GENERATOR VALVE

HEATERS ( l—’
GENERATOR

 

 

 

 

I
- 7 L——-TURBINE

LUBRICATION AND -
COOLING WATER—a | |

 

 

 

 

 

  

 

REACTOR
CORE

      
 
 

 

 

 

 

BOILER
PUMP e
PUMP

 

 

PRIMARY N
CONDENSER
COOLANT - fi ——  COOLANT

 

 

 

 

 

| COOLING WATER

 

CONDENSER

 

 

 

Fig. 2. Simplified Pressurized~Water Reactor With Hermetically
Sealed Primary and Secondary Systems.

Primary System

The basic heat genergtion,and removal system of a pressurized;water
reactor is relatively simple, but the auxiliary equipment associated with
it is often complex. The major complexities are in the reactor control
system, and, in some cases, the water~treatment system. It appears
possible, however, to satisfy the present control requirements with a
very simple system based on the inherent self-regulation characteristics
of a highly~enriched pressurized-water reactor. Water tfeatment does not
appear to be needed in a hermetically sealed primary system to be operated
for one year. The use of a hermetically sealed system with inherent
nuclear control reduces the pfimary system to essentially the core,

héat exchanger, coolant éirculating pump; and pressurizer. Various

‘featgres of the primary Systemfare discussed below.

 

":Reaétor Control

~ The action which controls the fissioning rate in a nuclear reactor

may be brokéh'doanifito thrég”functions. These are regulation, shim,

| andfsafety;r In a_préssuriZEdewatér:réactdrlthe regulation function

maintains the primary water at a constant temperature despite small

17

 
 

 

changes in reactivity, such as those associated with changes in load.

Shim control performs the same type of compensation for the largé're- \ ‘ )

activity changes associated with fuel depletion and accumulation of
fission~-product poisons. The nuclear safety system stops the chain re-
action in the event of malfunction of the reactor system. Interrelated
with these functions are the scheduled actions fihich start up and shut
down fihe reactor. '

The regulation function on a pressurized~water reactor is an inherent
one. The negative temperature coefficient is of sufficient magnitude to
control small changes in reactivity with only slight changes in the
average core temperature. Long~term changes in reactivity during the
life of a core may, however, be appreciable. The upper curve® of Fig. 3
illustrates the reactivity change which would occur in a fully=enriched

UNCLASSIFIED
ORNL-LR~-DWG 50108A

 

.30 ,

 

COMPUTED FOR PM-1 {REF &)
——e—e—— COMPUTED FOR SM-1 {REF 7}

\ NO BURNABLE
. NS POISON
s \

N

LUMPED BORON
'< INCREASING BORON

CONCENTRATION

 

1.25

 

 

 

 

k"f f

1.10

 

 

 

 

 

 

N\
TN
TN

“——UNIFORM BORON

 

 

 

 

 

 

 

0.9%
0 5 10 15 20 25 30

CORE BURNUP ( thermal Mw—yr)

Fig. 3. Illustration of Effect of Burnable Poison on Reactivity
Change of Pressurized-Water Reactor.

—

18

 

*
 

-l

pressurized-water reactor if there were no compensation by the motion
of a shim rod. The curve begins after the reactor has been heated from
room temperature to its operating level and after the initial reactivity
has been reduced by the accumulation of xenon. (The xenon poison
fraction reaches about 90% of its equilibrium value in one day.)

The addition of boron as a burnable poison serves to reduce the ex-
cess reactivity during the early part of the core life, as illustrated
in Fig. 3. The two lower solid curves® of Fig. 3 are estimates of the
effects on PM~1 of different amounts of lumped boron. The dashed curve?
shows the reactivity change with burnup in an SM=l=type core with uniformly
distributed boron; the analytical model used in computing this curve
accurately predicted reactivity changes during the life of Core I of SM=-l.
The dashed curve remains within a range of 1% Ak/k for a core life of
approximately 8 Mw~yr (thermal), and the lower solid curve remains within
that range for an even longer period. One year of operation of a reactor
producing 1 Mw of electricity would result in core burnup of 7 to & Mw=yr
(thermal ).

One difficulty with the use of a burnable poison is that it tends
to reduce the reactivity lifetime (the period during which keff exceeds
1.0) of the core. If a shorter reactivity life of the core can be ac-
cepted, curves even flatter than those in Fig. 3 can be obtained. Other
measures, such as using more than one absorbing element, closely tailoring
the poison location and concentration, and perhaps even tailoring the
fuelmlocation'and concentration, may extend the period in which the re-
aetivity remains'fairly censtant'- The uncertainties in the physics
calculatlons, however, become greater as the system becomes more complex
or as 1t deV1ates more from the reglon -of experlence. '

A reactor having 8 temperature coeff1c1ent of —2 x 10™% (Ak/k)/ F

would drop in temperaturerby 50°F to compensate for a 1% change in re=

' activity. (The temperature coefficient of the -SM-18 is more negative
— than -2 X 1074, ) If the 'burnable poison restrlcted the reactivity to a

change from.burnup of l%@ permitting the water temperature to vary over

a range of 50 F would ellmlnate the need for mechanical control durlng

19

 
 

 

routine operation., This technique appears to be feasible as a means of
obtaining high reliability for a period of one year.

The preceding discussion has been based on the use of fully enriched
fuel. Use of low=enrichment uranium might appear attractive. in a
partially enriched element an excursion is curtailed by the Doppler
effect in the fuel, whereas in fully enriched elements there is dependence
on the change in moderator density. Actually, a strong fuel temperature
coefficient is undesirable with regard to achievement of inherent con-
trol. The proven low-enrichment fuel elements for power reactors con-
tain bulk UO,, as discussed in the next section. The contact between
UO, and the cladding is such that there may be an appreciable temperature
drop between them, and the thermal conductivity‘of the gas in the gap
changes significantly upon the evolution of xenon and krypton (unless
the element is initially filled with xenon). In addition the effective
thermal conductivity of the UO, may possibly change with time because of
cracking and fuel burnup. The U0, temperature would increase during the
life of the core from these effects. The uncertainty in the temperature
change would make the increase in resonance absorption unpredictable, and
this would make it more difficult to limit the reactivity change to a
specified range. For this reason fully enriched uranium appears to be
the more attractive fuel for a reactor which does not have an automatic
control system. ,

It should be noted that the system proposed will be stable with
respect to xenon "transients." A small reduction in electrical load
would result in an initial decrease in neutron flux, followed by a
transitional increase in xenon poisoning. The increased poisoning
would lower the reactor temperature and reduce the temperature driving
force for heat transfer to the secondary system. This sequence would
continue if there were no corrective action. However, in a system with
a fixed electrical load, the thermal power would actually increase at
the lower reactor temperature, since the lowered thermal efficiency would
result in a greater heat load for the same electrical power. The flux

- would thus tend to increase and stabilize the system.

20
 

 

 

 

)

This inherent stability may not exist if, as discussed later, the
electrical load is automatically varied to control the frequency. With
control obtained by varying the electrical load, a reduction in reactor
temperature would be followed by a reduction in electrical load and
probably a reduction in the thermal load on the reactor. While a system
of this type might be stable, this can only be determined by study of
the particular case.

There are three areas of concern relative to a reactor with inherent
control: (1) the effect of core design on the temperature coefficient,
(2) the uncertainties in judging the proper fuel and poison loading, and
(3) the difficulty of accurately controlling the boron loading in a fuel
element. Changes in size, shape, fuel~-to=moderator ratio, reflector
thickness, poison concentration, etc., may change the temperature co-
efficient of reactivity. In the design of a self=-regulating core,
achieving a favorable temperature coefficient is as important as limiting
reactivity changes with time, and it should be studied as thoroughly.

The problem of estimating reactivity may, for convenience, be
separated into two parts, estimation of the initial critical mass and
estimation of the change in reactivity with burnup. The first part in-
volves an attempt to make the reactivity at the beginning of unattended
operation equal 1.0 at the temperature desired; the second involves
estimation of changes which are in a range of sbout 0.0l Ak/k. A dis-
tinction is made between these two parts of the problem because the
cause and effect of errors is different for each. For example, a 3%»
error in the initial reactivitycmight mean a 60°F error in water
temperature if not corrected, whereas a 3% error in the change in Ak/k
would have an in51gn1f1cant effect on operation of the reactor. A mis-
calculatlon of the first type is determlnable, however, and probably
correctable before operation of the reactor. An error of the second type
would not be known- until a core had been operated

,There;ls—en appreciable body of experience with cores of the general
SM-1 type that would be of value in estimating the initial reactivity,

andtcritical experiments.could'bejperformed if needed. 1In estimating

21

 
 

 

 

- reactivity changes with burnup, the calculational technique uSed would
be normalized against any experimental data that are applicable. There (:;
has been little experience in this area, however, and there are uncer-
 tainties associated with the calculations.? An extensive physics pro=-
gram is clearly called for, and it would be well if the caléfilatibns.
could be checked against an experiment early in the program. A particus-
larly useful approach would be to construct a core designed for inherent
control and install it in an existing reactor of the same general type.
There will be several suitable small reactors in operation by the:time &
core could be ready for testing. In any case, physics problems are not
likelyrto delay achievement of a reactor with inherent control,'since a
core with characteristics similar to those of the dashed line in Fig. 3
could be used if a core having better reactivity characteristics were |
not achieved.

The preceding discussion of burnable poisons has beenrconcernéd
with the calculation of the amount of poison required. Another problem
1s that of insuring that the desired amount of burnable poison is in=-
cluded in the fuel element. This 1s discussed later under the section
on fuel elements.

Even if burnable poisons are used for restricting long-term reactivity
changes, some method must be provided for insuring that the core is sub-
critical before it is brought to the operating temperature. In addition,
compensation must be made for the reactivity change which results from
the initial accumulation of reactor Poisons. Thus there must be a con-
trol system for bringing the reactor to the condition in which the
reactivity is on the flat part of the reactivity-burnup curve. Having
pbison rods perform this function during the startup period would not
-ieduce the reliability of the system, since the rods would not participate
_in the operation‘of the reactor when it is unattended. The system for
'providing controlled movement can be quite simple, since rapid insertion
would be under the force of gravity. |

~ There are absorber materials which have adequate reliability for

use in the control rods of an unattended reactor. Unclad hafnium would

:22

 
 

 

- )

satisfy the requirements, and it is likely that a matrix of europium
oxide in stainless steel would suffice. (Europium elements are being
tested in the SM-1 at present.) Other materials may also be satisfactory,
since, being normally withdrawn, the rods would undergo little burnup.

The inclusion of poison rods would be advantageous for another
reason. If late in the development program it were decided that the in-
herent control system were not desirable, or if the objectives for the
system changed, the rods would be available for shim control with no
appreciable change in reactor design. The rods could be used to change
reactivity during the 1life of the core if a means were provided for
moving them out at intervals by small amounts. For example, a watt=hour
meter located on the electrical output of the turbine could actuate a
simple mechanism for moving the rods. Alternatively, the rod movement
could be programmed in advance for a mechanism operated by a clock. This
type of control would appear to be much more reliable than one which
monitors the core temperature and attempts to compensate continuously
for changes, but, to operate as designed, all the rods would have to move
when directed. If communication with the reactor were possible, the out-
put of a thermocouple could tell an operator when to move the rods.

A fast=acting safety system does not appear to be required for a
low~power~density pressurized-water reactor, particularly in a remote
location. Probably the most hazardous time during the life of the reactor
is the startup period. During startup the system would be under the con~-
trol of an operator, and mEans could be provided for dropping the poison
rods on either an. automatiC-orfamanfialiéignal.'_Oncethe.reactor was in

operation, there would be little chance of a rapid sddition of reactivity.

Power removal would béISteady, gndlthére'WOuld be no pump startups or
' condifiion changes Which-might1prdvide the normal source of a cold~water

“accident.  Operator errbr7WOfiidlbe;éiiminated along with the operator.

In general when a water»cooled reactor is operatlng at a high power
1evel it is difficult to add react1V1ty at a rate which would endanger

the 1ntegr1ty of the system. ol _
- From the- precedlng discuss1on, there appears to be no need for a

fast—actlng safety system, except perhaps during startup. Some method

23

 
 

 

 

 

 

would have to be provided, however, for shutting down the reactor at a
signal ffom outside the system or at a specified time in the life of the
core. In addition, rod insertion would be prescribed upon failure of the
electrical output from the generator (since in any case this would signal
that the reactor was not operating as designed) and possibly upon high
température or high pressure. There is the posSibility of a spurious
signal causing an unnecessary shutdown, but a multiplicity of independent
circuits (perhaps a three-out—of-five'system) could be used to make the
.protective system more reliable., Aside from the automatic shutdown equip=-
ment, the system proposed would depend solely on the inherent properties
of the core for controlling the reactor when it was unattended.

The problem of removing fission~product decay energy has not been
considered, but the design should be such that there would be adequate
coolant flow by natural convection for afterheat removal. The wltimate
rejectioh of heat to the surroundings would depend on the specific

application of the reactor.

Ffiel Elements

There are satisfactory fuel elements for use at the power densities
and temperature levels envisioned for a reactor of the type proposed.
Core I of the SM~1l, for example, appears to have performed adequately
for over two and one~half years with a burnup of almost twice that re=
quired for the present application. Although SM~1 elements with full
exposure have not yet been examined in a hot cell, their behavior (and
examination at the reactor site) indicates that there have been no failures.

A number of cores of the same general type (i.e., flat plates of
fully'enriched U0, in a stainless steel matrix cled with stainless steel)
have been built, and there is an established manufacturing capability
for.this design. While the SM~l design would be quite adequate fbr‘the
present application, improved elements may result.from changes in fabri-
cation methods and from modifications, such as the use of spherical oxide
particles.

Elements with bulk U0, contained in stainless~steel tubes also
appear to give reliable performance, as indicated by exhaustive tests

24

ul
 

 

 

«3

and by experience with the PWR. Bulk U0, elements, however, are normally
only partially enriched with U?3%, As discussed sbove under the heading
Reactor Control, partial enrichment is less attractive than full enrich-
ment for a reactor with inherent control. Other types of fuel element,
such as plates of Zircaloy=-2~clad zirconiumpurahium alloys, or the
tubular stainless steel—U0, matrix elements developed by Martin for the
PM~1, might be satisfactory. From a reliabllity viewpoint, however,
there appears to be no necessity for use of other than the proven SMwl
type of element. .

There has been a problem with the inclusion of uniformly distributed
boron in SM~-l-type fuel eléments. During fabrication of the SM~l core,
a large fraction of the boron was lost from the fuel. Research con-
ducted during the past two years has, however, revealed the mechanisms
by which boron is lost and pointed the way to achieving better control
over the final concentration. The boron loss during sintering of the
stainless steel—U0, meat can now be closely regulated, but the loss on
fabrication of the fuel plate is less controllable or predictable. At
present, the final boron content of a fuel plate would probably be within
10% of the specified value. Work is continuing on this problem, and
improvement in control of the boron content of the fuel element is to be
expected. (For lumped poison, quite accurate control of the boron con-
tent could be achieved by adding machined strips or rods of boron=
containing materials. )

The desirability of using poisons other than boron would be deter~
mined by the reactivity advantage to be obtained as baianced against
uncertainties associated with their physical inclusion in the reactor.

It might appear desirable to use other fuels, for example, plutonium

- isotopes,?® to assist in the control of reactivity during the life of

the core. ' The technology:of including plutcnium in a fuel element has
not been devéloped nearly so far as that of uranium, however, and it is

therefore advisable at present'to use only uranium fuel.

25

 
 

 

 

Primary Coolant Pumps

As was discussed above, one advantage of a boiling-water reactor
would be the elimination of the primary circulating pumps. Alternatively,
a naturale-circulation pressurized-water system might be used. However,
it appears that experience with canned=rotor pumps of the size needed for
a 1l.0=Mw (electriéal) reactor is sufficiently favorable that not much
advantage would be gaihéd by‘eliminating them, particularly if their
‘elimination would mean going to a system with which there has been no
‘experience. Although one may point to many examples of failures of large
canned~rotor pumps, a detailed examination will show that most of the
 failures either occurred early in the life of the pump, &8 a result of
manufacturing defects, or they were a consequence of malfunctioning of
other parts of the system. Once a particular canned-rotor pump has been
checked and found to be good, there is a high probability of its lasting
a year and possibly even well heyond a year. As an example, both of the
core circulating pumps of the SM=-1 have operated without failure for a
period of two and one~half years in which they have been subjected to
many starts and stops (on one occasion a circuit bresker tripped without
discernible cause).11 In order to insure that there are no defeéts in
design and construction of the particular pumps to be used, they should
be tested in a loop and then run for a period during the checkout of the
assembled reactor.

For added reliability, it would be poésible to employ two continuously
operating pumps in parallel, with each having sufficiént capacity so that
if one failed the flow would still be high enough to continue to cool
the core. BSome type of simple check valve would be used to avoid by-
passing of the core by recirculation through an inoperative leg; Since
: the pressure drop would be low and a perfect seal would not be needed,
it would not be difficult to design a valve to give the needed relig~
bility. The procedure when switching from one pump to another in the
SM=1 is fo start the second pump, so as to have both operating, and then
to cut off the first pump. There have been no difficulties associated

with this method of operation,t?

26

G

 
 

 

 

)

i\

In the case of the SM=l, one pump will produce a flow of about 3900
gpm through the core,'and if both pumps are operating, the flow is 4600
gpm.l1 Hence, in this system, the flow would decrease only about 15%

upon the failure of one pump. A careful selection of pump characteristics

and operating points may result in an even smaller flow change, although

the above appears to be satisfactory.

Core Pressurizer

The pressurizer in a reactor primary system keeps the pressure high
enovgh to prevent boiling in the core. This function should be retained
in the reactor design, since there has been little experience with either
local (subcooled) or bulk boiling in pressurized-water systems, and
reactor control would be more difficult if boiling were permitted. In
addition to preventing boiling in the system, the pressurizer normally
acts as a surge tank to accommodate changes in the volume of the primary
coolant. The pressurizer will have to be designed to handle appreciable
volume changes in a system where water temperature variation is used for
reactor control. (A 50°F temperature change causes about a 5% change in
liquid volume.)

While there have been designs for self-pressurized reactors (which
allow boiling near the core exit) and for pressurizers using nuclear
heat, the electricaily heated pressurizer should be retained because of
its proven capability. The major limitation on the reliability of
pressurizers in existing systems is that associated with the device
which controls thé préésure; Normally'an'instrument senses the pressure
in the system:and uses a COntroi}tO'turnrthe heaters off and on.

The relisbility of _the,pr’es_«.éi_;rizer system would be increased if the

necessity forfsensing-the,SYStemfiiréssuie and for switching the heaters

' bnfand off'could'bé'eliminated,.;A Systemgfihafi”requirés no control or
_Sensing of preSsurefhaswbeen devised to_pérmit'continuous operation of
_ithe'electricairhéaters,f'?or=¢bfiétant cbfiditions_in the_primary'system,

the preSSurizerLhéatéré can be caréfullesiZéd'SOIthatlatieqpilibrium-

conditions'tfiéftemperature of the pressurizer is at the design level.

27

 
 

 

 

 

Such & heat balance is obviouSly precarious in that the temperature of
thé-pressurizer is a function both of the electrical heat input and the
heat transfer coefficient., In the system described, the temperature _

\ différenée between the pressurizer and ambient might be 400°F, and thus

a 10% change in heat generation or heat transfer coefficient would result
in a témperature change of 40°F (equivalent to perhaps 400 psi).

If the temperature drop from the pressurizer to the heat sink were
small (say 50°F), the presumed 10% change in temperature differential.
would only amount to a few degrees. By insulating the pressurizer from
the ambient and providing within the pressurizer-a heat sink cooled by .
the exit water from the reactor core,rthe pressurizer~to-sink temperature
differential and thus the absolute change in pressurizer temperature re-
qpired-to,accommpdate small changes in heat transfer cofiditioné can be
 sharply'reduced. -Since it is the temperature difference between core
and pressurizer that prevents boiling, the control of this difference
wlll produce satisfactory operation.

A pressurizer design based on the above principle of maintaining
the pressurizer temperature a specific amount above the core exit water
temperature by using a fixed heat input would require little developmentsl
work, and the system should operate reliably over the life of the reactor.
A system of the type described is illustrated in Fig. 4. The heat input
to the water in the pressurizer is provided through a number of parallel
electric heaters, each operating independently of the other and separately
fuéed, if necessary. If there were 20 independent heaters, loss of the
heat output of several would not jeopardize operation of the reactor.

Steam generated by the heaters would be condensed on tubes cooled
with core~outlet water. If the condenser tube wall were deliberately
made thick, the heat transfer resistance could be concentrated in the
wall and would not be sensitive to changes in the heat~transfer coefficient
on elther side of the tube. One problem of this particular design is
the interference of hydrogen (which would be present in the pressurizer)
with the condensation process. If this appears to be a serious problem,

it may be possible to keep the hydrogen from accumulating by connecting

28
 

 

 

 

*}

UNCLASSIFIED
ORNL-LR-DWG 501094

INSULATION

  
  
  
   
 

‘ PRESSURIZER
PRESSURIZER
COOLING
CIRCUIT

ELECTRIC
HEATERS

 

}—= TO STEAM GENERATOR

 

REACTOR CORE

 

 

 

 

 

~-——FROM STEAM GENERATOR
PRIMARY COOLING PUMP

Fig. 4. A Self-Regulating Pressurizer.

a small vent line from the top of the pressurizer back to some lower=-
pressure point in the primary system.

There are variations of this concept which perhaps would work as
well or better than the one illustrated in Fig. 4. One possibility is
to provide for the vapor generated in the pressurizer to condense directly
on a free surface of water from the core. Another is to locate a heat-
removal coil in the liquid volume of the pressurizer rather than in the
gas space. Heat transfer from the electric heaters to the coolant would
be by natural circulation of the water in the pressurizer. Operation

would not depend upon condensation and would not be affected by gas

'_accumulation, but the natural-convectlon coeffic1ents mlght be sensitive

to changes in'water level 1n the"pressurlzer One advantage of having

' 'lboth elements in the water Would be that the natural-convectlon coefficient
jdecreases as the heat transfer rate falls off and thus tends to stabalize
“iithe temperature dlfference between the pressurlzer and ‘the prlmary system.
_’_In contrast the condensing coeff1C1ent tends to 1ncrease as the heat
’tftransfer falls off and thus tends to reduce the temperature dlfference

),—faster than the heat generatlon rate decreases

It would appear that any of these de51gns could be made to work

qulte rellably. However, if attempts to avold pressurizer control are

29

 
 

 

 

not successful, an automatic control system using duplication or coinciw

dence circuitry could be made to be reliable. ‘ }

 

Main Heat Exchanger

Since the heat exchanger has no moving parts and no control equip~
ment, for this unit, reliability is synonymous with leaktightness,f In
order to assure lesktightness, not only initially but for the desired
life, the following procedure would be followed:
1. A material would be selected that is easily fabricated, has good
~corrosion resistance, and is not subject to stress~corrosion cracking.
Inconel appears to be such a material. Its use is discussed more
thoroughly under the heading Water Treatment.
2. The héat exchanger would be carefully designed with particular

emphasis on the tube~to~header joints. Any conflict between economy and
- reliability would be resolved in favor of reliability. '

3. There would be close inspection throughout fabrication.

4. The completed exchanger would be thoroughly checked for leaks.

5. The heat exchanger would be installed in a loop and tested at
reactor conditions for an extended period. _

6. After loop operation, the exchanger would again be inspected
and tested for leaktightness.

A heat exchanger constructed and tested as described should have a

very high reliability for many years of operation.

Water Treatment

 

The presence of excess hydrogen in the primary system is desira-
blel2=14 principally because the radiolytic oxygen content is then held
at very low'(often undetectable) levels by the radiation~induced water=

recombination reaction

2H, + 0, — 2H,0 | (3)

30

 
 

 

 

"

W

Owing to the sealed condition of the system, this‘excess hydrogen cannot
be consumedrby atmospheric oxygen and nitrogen (which normally would be
introduced with makeup water), and it will not be lost by leakage, except
possibly by diffusion through the contaliner walls. Additional hydrogen
will be produced by the overw~all corrosion reactions of metals such as

iron and chromium; e.g.,
3Fe + 4H,0 = FE3O4 + 4H, (4)

If it is assumed that the primary system is principally stainless steel
and that the sverage corrosion rate is 5 mg/am?'mo,15 it is estimated

that approximately 30 standard liters (1 ft°) of hydrogen will be produced
per 100 ft2 of surface in one year of operation. The resulting gaseous
hydrogen pressure in the pressurizer (of the order of a few psi) would

not be excessive.

Other primary water treatment methods used in conventional pressurized=-
water reactors are pH control (usually in the range 9 to 11) and side-
stream water cleanup by filtration and/or mixed~bed ion exchange,12-13:16
The need for these water-treatment procedures in the present system will
be determined mainly by the degree of heat-transfer surface fouling by
transportable corrosion-produce solids (crud).

pH Control. It appears that in stainless-steel and carbon-steel
primary systems, 12"'13’1“‘"":“3:high pH (9 to 11) has a distinetly beneficial
effect in reduc1ng “the. quantity of corrosion products released to the

coolant Recent results17 1ndlcate that the release rate of corros1on

iproducts from carbon steel at a pH of 10 (~l mg/am -mo,at 450°F) is of the
~_order of eight tlmes less than the release rate at a pH,of 7 Release |
_ _i_jf;irates for stalnless steel and Inconel at a pH of lO to lO 5 and at 500°F
-"::Lgswere < mg/fim? mo and.~4 mg/fim?»mo, respectlvely, after 200 hr'of exposure.
fséuiln.both cases, the results 1nd1cated that lower release rates would be- found:f-:
"ifiliupon longer exposure If it is assumed as the worst case that all corrosion
flfsolids released to the coolant are dePOS1ted on the fuel-element surfaces,

'fftan average release rate of 2 mg/dm mo 1n a stainless steel-Inconel

31

 
 

 

 

 

primary system (pH of 10 to 10.5) in which the fuel-element surface is
one-fifth the total area would result in a fouling rate of 10 mg/am?-mo.' &i;
In a year of operation this would give a deposit averaging ~0.00017 in.

in thickness. On the same basis, at a neutral pH, a deposit of the order

of 0.001 in. in thickness (estimated from the effect of pH on the release
rate from carbon steel) would result.

Thus it seems that a basic pH in the primary system is desirable,
even though the over~-all fouling problem is not so severe that it would
be disabling in a system with moderate heat flux. The pH of the coolant
could be established at the outset by the addition of lithium hydroxide
or ammonia. The stability of pH with time (if ion~exchange control is
'_'absent).in such a sealed system is difficult to predicf; however, since
there are no base-consuming processes at once apparent, it 1s expected
that the pH will remain sufficiently above the neutral point to be dis-
tinctly beneficial from the standpoint of fouling.

Cleanup. Side~stream cleanup by mixed~bed ion-exchange resins in
conventional pressurized-water systems has three functions: (1) to re-
move soluble water-borne activity, (2) to filter out insoluble water-
borne corrosion solids in order to reduce crud deposition, and (3) to
control pH by introducing the cation~exchange resin in the lithium,
ammonium, or hydrogen form. In the present system, cleanup of water-~
borne activity is not a primary consideration. Removal of water-~borne
crud probably will not be an important consideration in view of the ex~
pected low fouling rates at high pH, especially since side-stream processing
may not greatly reduce fouling in any case.l? Finally, the added pH
control afforded by a mixed-bed ion exchanger will be needed only if some
unforeseen base~consuming process occurs in the system. Thus it seems
gquite possible that a side~stream ion exchanger can be omitted.

Since the crud~removal rate could be greater, an on~stream cleanup
system would offer promise of greater reduction in fouling than a side-
stream cleanup system. BSuch a system, however, must in no way endanger .
reliable operation of the primary system. A means of such cleanup could

be based on the fact that the principal constituents of the erud will be

32
 

o s e e e e o

 

 

 

o

magnetic oxides of iron and chromium.}?® It seems possible that on~stream
removal of crud, if needed, could be accomplished by using a magnetic
collector in a region of low flow rate.

To summsrize this discussion of water treatment, it appears likely
that a hermetically sesled'primary system would operate satisfactorily
for a year without any treatment other than the establishment of desirable
initial conditions. The preferable initial condition appears to be the
use of de~ionized water which 1s adjusted té a high pH and whiéh contains
dissolved hydrogen. A research program involving high-pressure loops
would help determine the conditions which are favorable for operation
without water treatment. Operation of existing pressurized-water reactors
without continuous purification might provide valuable data on crud
transport and deposition. If further study indicates that side~stream
purification (or possibly on-stream-cleanup by magnetic collection) is

desirable, its provision would have little effect on system reliability.

Secondary System

 

The decision to employ a conventional steam secondary system for
power generation was based on the high degree to which such systems have
been developed. It was recognized, however, that the conventional steam
cycle will have to be simplified and improved if the ability to operate
unattended for one year is to be achieved.

Three concepts of the steam system are considered in this study:

(1) a "simplified conventionai ‘steam cycle from which feedwater heaters,
hot~well ‘pump, b01ler blowdown, and venting of noncondensibles are elimi~

nated; (2) a. semiconventiona ‘-system whlch has the preceding simplifi-
cations plus a moisture-recovery system that avoids net loss of water from

the turbine shaft seal, and (3) a'hermetically sealed steam system in

- whlch‘water loss from the system is positively ellmlnated Each of these

'concepts W1ll be dlscussed in turn.

33

 
 

 

Simplified Conventional System

Feedwater heaters are excluded from the power system, since thermal
refficiency is of secondary importance. The hot-well pump is elimifiated
by using a multistage boiler feed pump to perform both pumping fuhétions.
By selection of a pump which is sensitive to suction head, the level in |
the hot well, and, consequently, the level in the steam generator, can be
controlled automatically.

Boiler blowdown can be made unnecessary for one year of operation by
using a material such as Inconel for the steam generator and pretreating
the water supplied to the system. With the elimination of blowdown,
makeup water is needed only to replace that lost from the shaft seals.

A water cleanup system does not appear to be necessary, although its pro-
vision in a side stream would not reduce the'reliability of the power
plant. EJjection of noncondensible gases may not be required, since the
amount of gas accumulated in one year is estimated to be small enough
that it could be accommodated by proper condenser design.

A brushless type of generator would eliminate the frequent maintenance
required for brushes and commutators. The turbine admission valve would
be the only valve required to operate after startup of the plant. The
system could be of all~welded construction, with bellows~sealed valves.
It would be leaktight everywhere except at the shaft'seal, and water
losses from the shaft seal could be accommodated by having a large water
capacity in the system or by providing a makeup storage tank.

The steam pressure would be relatively low, perhaps 300 or 400 psia,
and no superheating would be required. With proper turbine design the
steam path would tolerate the higher moisture content of the low-pressure
stages. Since efficiency is not of major importance, a relatively high
condensing pressure could be used if made desirable by the nature of the -
heat sink and by’ moisture considerations in the turbine.

The eduipment required for this cycle is all within the realm of
present technology and can be manufactured with little or no development
required. The need to replace lost water seriously handicaps this type
of system, however, and precludes its recommendation for the particular

application now under study.

34
 

 

 

 

&

Semiconventional Steam System

The simplified cycle described above may be further modified to
eliminate the net loss of water by recovering the leaskage from the tur-
bine shaft seal and returning it to the system. Several schemes that
were studied were considered to be workable. Perhaps the most practical
of these is the use of a conventional bleed~off type of labyrinth seal
that is vented to the main condenser, as illustrated in Fig. 5. The seal
would separate the steam at the tail end of the turbine from the gas in
the containment vessel. The vessel could be initially filled with inert
gas at a pressure greater than that in the condenser so that a net in-
leakage to the system would occur. Noncondensible accumulations in the
condenser would be withdrawn by a steam~jet ejector, passed over an
after~condenser to reduce the moisture content, and discharged back into
the containment vessel. Once equilibrium conditions were established,
no net water loss from thevsystem would result unless there were surfaces
below the dew~point temperature in the containment space. If condensation
were difficult to avoid, a cold surface could be provided with means for

returning the condensate to the condenser through a hydrostatic leg.

UNCLASSIFIED
ORNL-LR-DWG 501104

. HIGH-PRESSURE
//’7 STEAM CONTNNMENTVESSEL’//IV ‘\\

 

 

 

 

 

 
 
  

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

SEAL
BRUSHLESS
GENERATOR
TURBINE
R li;,ff7' T _y“,,/AFflH?CONDENSER
_LO_WTZPR_ESSURE_ STgAM || st \__/__ T e
. S iy e ;EQ?CTOR | AFTER CONDENSATE
‘ QONDENSER  o ' :
I o FRaE — . GAS AND VAPOR
Tl e LEAKAGE ‘TO SEAL
\ . TOFEED PUMP "

-/_

_Fig;}sf,Séai;f6r78ifiplifiédConvéntiénaluSfieamzsystem.

 

35

 
 

 

 

This concept permits use of an essentially standard turbine with a
brushless-generator set and, yet, largely elininates the makeup water
problem. The need for a steém ejectdr and after~condenser, however, adds
complications to the system'which'are preférdbly avoided; thus éonsidera—

tion was given to a completely sealed system.

Hermetically Sealed Steam System

 

By using a canned turbine~generator to eliminate the shéft seals
from the otherwise leaktight semiconventional plant, a hermetically
sealed system which eliminates the water leakage problem can be achleved.
The rotor of the generator would operate in a water or water~vapor atmos-
phere and would be cooled by circulation of cooling water through the
stator. Generator manufacturers and manufacturers of canned-rotor pumps
state that generators of this type and of the requisite size can be
fabricated using presently existing technology.

Operation of the turbine~generator as a sealed unit mskes it desirable
to use water~-lubricated bearings and to eliminate an oil~actuated
hydraulic-governing system. Turbines with water-lubricated bearings
have been tested by several manufacturers, and water-lubricated bearings
have been quite successful in canned-rotor pumps. There has been little
experience with either water-actuated hydraulic governors or water-
lubricated mechanical governors, but there appears to be no major diffi-
culty in developing such equipment. Both the governor and the admission
valve could be located in the turbine housing. Other alternatives to
conventional~speed governors could be based on electrical sensing of the
generated frequency.

Boiler feedwater could be used for bearing lubrication and generator
cooling, since it would be cool and at high pressure. As shown in Fig. 2,
it could also be used for cooling the primary circulating ?umps.

‘While the hermetically sealed. system poses advanced design problems
that require more development work than would be required for the two
' concepts previously described, the extension of present technology is not

great. The hermetic seal concept promises a system of maximum simplicity

36

Th
 

 

 

 

q;)

0

£

m

with a minimum number of components. A further consideration is that,
should a design and development program on this type of system meet un-
expected difficulties, the effort could be diverted to the controlled-
leakage concept with little delay to the over=~all program. The components

included in the hermetic seal concept were shown in Fig. 2.

Basic Features of the Steam Cycle

Special problems are associated with the achievement of reliability
of some major components of the proposed steam system. These are dis-
cussed in this section. In addition, & discussion is included on methods
of handling control functions and of eliminating the water~treatment
problem.

Turbine~Generator. In order to assess the degree of reliability to

 

" be expected from conventional turbine~generators under normal conditionms,

Myers?® undertook a study of the operating history of seven units of 20~
to 25-Mw rated capacity. The operating history for these units, located
at the Oak Ridge Gaseous Diffusion Plant, covers approximately & l4-year
period. The units are conventional and are operated in a manner normal
for power plants. Myers concluded that the probability of one of the
units operating continuously for one year was only 0.05. In an extension
of this study by the authors, the operating histories of three units of
3-Mw capacity were studied, and approximately the same order of relia=-
bility was obtained. (It should be noted that the units referred to in
the above studies were not constructed to other than standard specifi-
cations. In addition, these particular units were built during World
War 11 and conseqnently are not only of an older d851gn but undoubtedly
possess some compromises in materials of construction )

A study of the maintenance work performed on the 3«Mw turbine-

generator units: 1ndicated that replacement of the commutator brushes on

';the ex01ter constituted the single most freqpent cause of outages. A

“close second in repair frequency was repacking of the turbine admission

valve In order to determine the improvement that might be realized by

the elimination of certain repairs,ra tabulation was prepared in which

37

 
 

 

 

 

 

it was assumed that valve-packing maintenance and exciter-brush replacement
could be eliminated. The elimination of these two repair items increased gfi;
.the probability of one year of successful operation from 0.05 to 0.60.

These data may be somewhat misleading, since minor adjustments or
repairs are often made without shutdown of the unit; there are wide 7
differences in opinion as to the effect of repairs made during operation.

Such uncertainties in the meaning of operating data limit the value of

statistical information on component performance. Neverfheless, an |

examination of the faults which caused forced outages of the units showed S
‘that the great majority were concerned with turbine auxiliary equipment

and were not outages resulting from failures of basic components in the
turbine~genersator. |

Although it is not uncommon for turbine~generator units to operate
continuously for a year without shutdown, experience in five nuclear
plants, as reported by Gilbert Associates,l indicates that 30% of all
forced station outages are due to the turbine~generator unit. Of these
outages, the majority resulted from the exercise of protective controls
to prevent equipment damage. Many of the other shutdowns were concerned
with repairs to the brush and commutator systems.

It appears that with the precautions of conservative design and the
elimination of the brush problem, an oil~lubricated turbine~-generator
unit having a high probability of successful operation for one year.can
be developed. As discussed elsewhere in this report, however, a water-
lubricated turbine-generator suitable for use with the hermetic seal
concept is a preferred design and can be developed in the allotted time.

Although it is probable that the brush life experienced on conven=- ' -
tlonal generator units can be extended by improved design, the predictable
reliability of brushes is insufficient to permit their use in an un~
attended plant. The use of slip rings might prove to be satisfactory,
but generators of the brushless design would be preferable, and there
is sufficient technology and experience with brushless generators to
warrant their consideration. Three types of brushless generators —

induction, rotating-rectifier, and inductor — are discussed below:

38 - ~

1]
 

 

 

 

e, ‘
o,

A\

1. Induction Generator. An induction motor may be operated as an

 

induction generator by driving it above synchronous speed to obtain
"negative slip." The operation of such a system on a network having a
lagging power factor requires the use of capacitance in parallel with
the load to obtain a leading power factor. A system of this nature has
inherently poor voltage control under conditions of varying reactive
load. A change in load or power factor causes a change in the excltation
current, which, in turn, changes the output voltage and frequency. To
correct for this shift, capacitance may be switched in and out of the
system as a function of load changes. For systems with a relatively
constant load, it is possible to operate with a fixed capacitance and
still maintain reasonable voltage regulation. Thus load swings of 10%
on a system with a unity power factor have been estimated to change the
voltage less than 2% of full=load voltage. External excitation may be
required during startup of an induction generator.

Induction generator units in sizes smaller than 1 Mw have been
constructed and cperated successfully, and there has been extensive
experience with canned induction motors on primary pumps. Banks of
capacitors operated in parallel with the load should be inherently long-
lived and may be designed to utilize a large number of small fused units
so that failure of any individual capacitor has little effect on the
system capacitance.

2. Rotating-Rectifier Generator. A brushless generator of the

 

synchronous type is also available. ‘The exciter for such a machine 1is

located on a shaft extension of the generator The armature of the ex~

citer is fitted with hermetically sealed rectifiers (also rotating with

~ the shsft) to convert the a-c output to d-c current for the main generator
‘field w1ndings. The rotating rectifier thus eliminates the need for

‘ commutator brushes;

The silicon rectifiers used in this type of . generator have good

rectiflcation efficiency, are capable of operating at relatively high
;temperatures, have high v1bration and centrlfugal ratings, and have been

used extensively and successfully. They are subject to rediation damage,

39

 
 

 

 

 

but 1t should not be difficult to proteét them with shielding. Synchronous

- exciters with rotating rectifiers have been built in sizes up to approxi—' , (;;
mately 200 kw, and, currently, at least one exciter of 1300-kw capacity

“is under construction.

3. Inductor Generator. Another brushless type of generator con-

 

sidered for this application is the inductor generator. Excitation is

obtained from the stator windings. A series of staggered "teeth" on the

rotor provide the path for the main pole flux, and alternating voltage

is induced in the main stator windings by variation in the reluctance of %
the air gap. Rectifiers supply the d-c current to the excitation statof
windings. Inductor generators have been built for frequenéies in the

range of 1 000 to iO 000 cps. The output voltage wave from an inductor
-generator will be influenced by a greater percentage of the higher harmonic
components than is normally present in a synchronous machine, and this

" could affect the operation of other equipment in the system. Manufacturing
experience with this type of machine, particularly for three-phase appli=-
cation and for lower frequencies, is somewhat limited. There is therefore
some hesitance in recommending it for this application.

The operation of either an induction generator or a synchronous
generator appears to be entirely feasible for one year of unattended
operation. Generator efficiency for these designs is estimated to be
above 80%.

An important additional consideration in the discussion of turbine=~

generators is the method of bearing lubrication. Conventional turbine~ &
generator units employ oil lubrication, and the semiconventional system
previously discussed does not preclude the use of an oil system. While -

such a system normally implies oil pumps, oil filters, and an oil cooler,
the oil pump could be directly driven off the turbine shaft, and self-
cleaning oversized filters could be installed in parallel to preclude
flow stoppage due to clogging. As pointed out by Burns and.Roe,zo much
of the complexity of a circulating~o0il system might be avoided with
properly designed bearings using oil reservoirs and multiple rings to

eliminate the need for forced~oil lubrication. Oil slingers or

40 »
 

 

A

“

!

L1

deflecting vanes could be mounted on the shaft to prevent oil leakage,
and cooling could be provided in the oil reservoilr.

The operation of conventional turbine-generator units has demon=-
strated that the interchange of bil and water in the systems is essentially
negligible, particularly if oil temperatures are high enough to prevent
moisture condensation. Oil wiper rings effectively reduce leakage of the
0il around the shaft. Water~vapor leakage to the oil system may be pre-
vented by.the use of centrifugal oil seals or other techniques. Thus
oil-water vapor interchange will not prove a limiting factor for one
year of operation.

An investigation was made of the possibility of using water-lubricated
bearings for the turbine~generstor unit. Water~lubricated bearings have
been employed successfully in canned~-motor pumps, and some units have had
motor capacities in excess of 1000 hp. Recently, steam~turbine units
with water bearings have been built and operated successfully. Units oft
up to approximately '750~hp capacity are currently under test and in
operation,21=22

Water bearings use hard-surface materials, such as aluminum oxide,
and thus are resistant to scoring by impurities. Bearings of both hydro-
static and hydrodynamic design have heen tested. Water temperétures
have been in the order of 100°F, but somewhat higher temperatures are
considered to be acceptable, and widely varying water supply pressures
have been employed. |

In order to assess the probability that water-~lubricated bearings
could be successfully developed and applied to turbine-generator units
within the scope of the ground rules for this study, discussions were
held with manufacturers experienced with the operstion of steam turbines

designedfwith,water-lubricatédfbearings.: Discussions were also held

‘with manufacturers of_cannedebtor.pumps employing water-lubricated

‘bearings. It was the unanimous opinion of these consulted that water

bearings'suitable'fOr'thiS~applicatioh could be developed within the
time’reqpired and that this development did not represent any major

extrapblatibn of preSent_technology;"

41

 
 

 

 

 

Direct coupling of turbine to génerator is probably preferable to
the use of a geared unit, since reduction gearing would introduce an
additional component that would be subject to failure. The direct
coupling,.however, requires a compfomise between optimum turbine speed
and optimum generator speed. | i '

It is desirable that the steam flow path in the turbine be conserva-
tive in design velocities. Turbine clearances may be larger than in
current practice to insure against rubbing failures, although'a small -
loss in efficiency will result. Protective coatings can be used on the
turbine blades where the moisture éontent of the steam isrhigh; '

Water Chemistry. One of the striking cOmfilexities of a conventional
steam secondary system is the equipment requifed for meintaining adequate
water purity in the operating system. It was felt to Dbe bf.fundamental
importance to the success of an unattended reactor plant that this portion
of the secondary system be simplified, and therefore a study was made of
the water chemistry of the secondary system. The study of this aspect
of operation of the steam power cycle included consideration of the
effects of corrosion products‘and the necessity for steam~generator
blowdown, the magnitude of water losses and makeup requirements, radio-
lytic gas formation and venting problems, use of chemical additives, and
means of cleaning up the circulating stream.

1. Water Treatment, Makeup, and Blowdown. A conventional steam

 

plant incorporates either a continuous or periodic blowdown of the solids
from the steam generator. Makeup water is supplied to replace water lost
in blowdown, from vents, and in system leakage. Since the unattended
reactor installation must be entirely self~-contained, the need for sub-
stantial quantities of mekeup water would require.either a large storage

facility, a water-recovery system, or a processing system for providing

- makeup water. The plant must be capable of one year of operation, and

therefore even nominal makeup requirements would assume large proportidns.

Furthermore, a system for supplying or processing makeup water normally
requires sensors and controls, and such components would decrease over~

all system reliability.

42

C

O
 

 

A}

L]

i

from the reactor flux will be due principally £o the radiation from

2. Corrosion. In conventional steam cycles the principal corrosive
conditions arise from the presence of oxygen and carbon dioxide in the
feedwater and from the leakage of these atmospheric gases into the con-
denser system.23 The usual treatments are deseration of the feedwater;
the addition of oxygen scavengers, such as hydrazine or sodium sulfite;
and the use of volatile basic compounds, such as ammonisa, organic amines,
or morpholine, to react with carbon dioxide and to increase the pH of
the condensate. In the present system, oxygen and carbon dioxide initially
present could be removed by purging and using suitable scavengers. If
radiolytic oxygen 1s not produced in appreciable amounts, water control
in a closed system should be considerably less difficult than in a con-
ventional system.

The operating reliability of the secondary system is critically
affected by the accumulation of corrosion-produce solids in the steam
generator. In conventional systems, corrosion~product production and
accumulations are minimized by maintaining a high pH in the boiler (with
caustic or basic phosphate salts) and in the condenser system (with
volatile amines) and often by the use of filming amines (10~18 carbon
alkyl amines). Accumulated solids are periodically removed by boiler
blowdown. Since blowdown is undesirable in the present system because
of makeup-water requirements, consideration was given to holding solids
accumulation in the steam generator to within acceptable limits by (1)
the choice of suitably corrosion~resistant materials, (2) the use of
permanent water additives, and (3) the use of demineralizers or other
water cleanup methods. ' ':i, , _'_ i'

3. Oxygen and Hydrogen Production _ The production rate of radio~-
lytic oxygen and hydrogen in the secondary system'W111 be much lower than

~in the- primary system, but thefrelative recombination rate probably will

ealso be less because these gases Wlll tend to accumulate in the vapor

spaces. The oxygen production rate in a steam generator well shielded
N16

in the primary water. For. an Nl6 activity ievel of 100 pe/ml (which has

‘been reported for the Idaho Test¢Facility),24 the gamms energy production

43

 
 

 

 

rate is approximately 3 X 107 Mev/sec-ml. If this were totally absorbed
by the secondary water [taking the G(0,) as 0.2],2° approximately

6 X 1010 molecules of oxygen would be produced per second per milliliter
of secondary water, If the back reaction due to recombination is ignored,
this could amount to an accumulation of several milliliters of oxygen per
kilogram of secondary water in a year.

This estimate is, of course, high, since not all the gamma energy
will be absorbed by the secondary water, and since the recombination rate
will become appreciable as the oxygen and hydrogen ejected to the steam
phase are recirculated in the sealed-system.26, In parfiicular, as excess
hydrogen builds up in the secondary system from the corrosion of iron
and chromium, recombination will occur more readily. Thus, with a well-
shielded steam generator, no éppreciable steady-state oxygen level is
expected; however, it is not known what oxygen level would result if the
steam generator were not shielded from the reactor. |

Assuming an iron oxide—chromium oxide production rate of 10 mg/dm?-mo
(which is reasonable for a stainless steel or Inconel system that incliudes
some carbon steel),27 it is calculated that the corresponding hydrogen
production rate will be 60 standard liters (2 ft?) per year per 100 f£t2
of hot water-steam surface area. This hydrogen will collect predominasntly
in the condenser vapor space, If unacceptably high hydrogen production
is anticipated, a void volume can be provided in fhe condenser for hydro-
gen gas accumulation. Alternatively, excess hydrogen might be filtered
off to a sepsrate tank through a palladium metal barrier.

4. Corrosion Solids Accumulation in the Steam Generator. The order

 

of magnitude of the corrosion=product accumulation in the steam generator
can be estimated from the corrosion rate of 10 mg/dm?-mo assumed above.
If there were 500 f£t2? of hot water and steam area, approximately 750 g
(~1.7 1b) of corrosion products would be produced in one year. A uniform
distribution of these corrosion products in a steam generator of 500~gal
capacity would reéult in a solids concentration of 400 ppm. This is com-
parable with the allowable suspended=solids concentration of 250.ppm set
by the American Boiler Manufacturer's Association?® for a 300- to 450-psi
 

 

"

)]

n

b

steam generator. If all solids were uniformly deposited on 200 ft? of

‘heat exchanger surface, this would give a scale approximately 0.006 in.

thick (400 mg/dm?). Since not all corrosion products will be transported )
to the steam'generator, and since they will not all be deposited on heat

transfer surfaces, these numbers suggest upper limits for good materials

of construction.

5. Permanent Additives. From the above rough calculation, it seems

 

quite possible that a sealed secondary system would operate successfully
for one year or longer without any water treatment at all. As in the
primary system, however, it also seems likely that the quantity of trans-
portable corrosion products could be reduced by increasing the pH of the
secondary water with a permanent additive, and this should be seriously
considered. |

Of the various pH additives available, the most promising from the
standpoint of long~term stability are (1) phosphate buffers, (2) caustic,
and (3) ammonia. The first two would provide a basic pH in the steam
generator The use of caustic, however, involves some risk of caustic
stress corrosion as a result of concentration by boiling. Ammonia would
provide a basic pH both in the steam generator and in the condenser. The
use of morpholine should also be considered. Like ammonia, it would
provide a basic pH in both the steam generator and the condenser, but
its long~term stability would require investigation.

In recent years, the use of long-chain alkyl amines (e.g.,
octadecylamine) as corrosion inhibitors in steam~plant condenser
systems has met with considerable Success.' It has been stated that
the residence time of the amine in the corrosion-inhibiting film is

only a few. hours, 29 and when used in conventional plants, amines are
“added continuously or intermittently At the same time, it is reported29
'that no- appreciable decomposition has been found in prolonged high- N
-temperature boiling tests in the laboratory Thus it may be possible

that the normal continuous addition of filming amines is necessary

'-1mainly because of leakage from the system.. If further 1nvestigation
_should show that filming amines have sufficient radiation resistance and

useful protective lives, they should e cons1dered

45

 
 

 

 

6. Cleanup. Since conventional boiler blowdown is not desirable
in the present system; alternative cleanup methods should be considered. | ' Qfi;
As pointed out above, large amounts of corrosion~product solids are not
anticipated, unless, of course, signifidant quantities of carbon steel
or other relatively corrosive alloys are used in the sécondary system,‘
A margin of safety can be provided in the design of the steam generator
by making provisions for solids accumulation in low parts. Possible
cleanup methods include side-steam filtration, side~stream evaporation,
and side~stream or on~stream magnetic collection. Limitations of each w
of these methods exist. ‘The filter may be subject to plugging (dépending
on the nature of the solids)j the evaporator system may require automatic -
control; and the magnetic collector would be effective only in removing
megnetic oxides. The need for recourse to these methods is not antici-
pated, however, unless relatively corrosive alldys are used in the
secondary system. |

From the foregoing it seems clear that operation of the secondary
system without continuous water treatment or blowdown is quite possible.
In addition, the use of permanent additives might reduce the quantity
of transportable corrosion products, and side-stream cleanup methods
could be employed if shown to be necessary. It is evident that investi-
gations of the problems associated with the secondary system water
chemistry should be initiated early in a program to design and construct
an unattended nuclear power plant.

Steam Power Cycle Controls. Instrumentation and controls for the

 

steam power cycle represent the major threat to reliability, according

to the survey made by Gilbert Associates.?!

As stated in their report, >
"Failures in this category [instrumentation and controls] have produced

more unscheduled losses of load than the total of all other categories.”

Clearly it is desirable to remove all unnecessary controls from the power

system. The elimination of a reactor control system is made possible by

shifting a portion of the control function from the primary system to

the secondary system. Thus changes in primary~system conditions, such

as temperature drifts due to reactivity changes, must be accommodated ‘

by the secondary system.

46 o -
 

 

 

 

W

»F

' L

 

-715F:1ng praperly

For a system in which the load is very nearly constant, the specified
frequency control to w1thin 1% is a loose requlrement Onrconventional
turbine~generator unlts, where much tighter frequency control is normally
exercised, hydraulic governor units act through hydraulic pilot valves to
operate the turbine admission valve. Operating experience with hydraulic
governor units is extensive, and such units are considered highly reliable.

If oil=lubricated bearings are employed on the turbine=~generator
unit, then a common oil system may be used with an oil~actuated hydraulic
governor. Experience indicates that oil leskage from the governor system
is negliglble, and, as was the case with oil-lubricated bearings, the
system can be designed to tolerate a small amount of leakage. It is
imperative that hydraulic governor systems have efficient filters to
remove solids from the hydraulic fluid.

Operation of én oil-actuated.governing system within a hermetically
sealed steam system is undesirable because of possible intermixing of
the oil and water. Use of water as the hydraulic fluid with an essentially
conventional hydraulic governor is an attractive possibility, since the
backlog of experience with hydraulic governors demonstrates their relia-
bility. Discussions were held with a manufacturer presently testing a
water=actuated governing system on a steam~turbine unit.?? The hydraulic
system being tested is of a force~balance type similar to that for an
oil=~actuated system. Pressure is supplied from small pump vanes located
on the turbine shaft. The degree of regulation achievable with this
system is comparable to that of - an . 01l-actuated governor, and, although
operatlng experience is limite&, 1t is the oplnion of the manufacturer

V,that the‘reliabillty 15 comparable to that of oil-actuated systems.

The water-actuated gcvernor currently in operatlon contains a backup

 'f;regulat0r that permits a second speed-regulatlon band. Such a de318n

,;'i;flww'be desirable for a reliable turbine-generator appllcatlon. ?TWD'

'i i€; fT1ndePendent governor systems, one set at 1/2% erQPency varlation ‘and
”"jiélthe other: at 1% frequency variation, could be emP10y9d | The wide~range

'7fsgovernor would be 1noperative if the narrow—range governor were function-

 

47

 
 

 

 

 

A second type of governor capable of operating within a hermetically
sealed turbine-generator unit is the mechanical governor. The preference
of the hydraulic governor over the mechanicgl governor for thé-majority of
present-day steam turbine-generator applications is based in part on the in-
creased precision of frequency control available with the hydraulic system.
The.rather loose requirement of 1% frequency variation with an essentially
constant load may permit the use of the mechanical governor for this appli~
catioh. The essential changes required for operation of a mechanical govers-
nor include éubstitution of materials to permit opération in & vapor environ-
ment. A difficulty encountered in operation of turbine-generator controls
is sticking of the turbine admission valve. The force required for valve
operation is an important consideration in the use of mechanical.governing
systems where the power amplification is somewhat limited. 'However, this
problem, which results from oxide formation on valve stems, has béen as-
sociated primarily with plants operating at high steam fiemperatures-and pres=
sures and should not exist at the temperatures anticipated for this system.

Other methods for achieving governor control in a hermetically sealed
system could be based on the electrical output of the generator. An
electrical governing system usually functions by sensing the frequency
output of the generator and comparing it with a desired set=-point frequency
to obtaln an error signal; the error signal operates a position controller
that regulates the turbine admission valve. A hydraulic or mechanical
governor external to the hermetically sealed system could operate off a
synchronous motor to regulate the turbine sdmission valve, and no penetra=
tions of the sealed system would be required.

An alternate scheme for eliminating the turbine governor and admission
valve was examined. This system would vary a dummy électrical load, in
parallel with the normal system load, to control the turbine speed. It
is envisioned that saturable reactors in series with a resistance load
would vary the impedance through the dummy circuit as a functlon of but- :
{put frequency. Solid-state components can be employed throughout the
circuit. Although this system has not been developed, the advantage of
completely eliminating the need for a turbine governor or turblne ad-
mission valve makes it of interest to the present application.

48
 

 

A¥

The exact type of frequency-governing system to be employed would
require a more detailed analysis than that presented here. It appears,
however, that several schemes are capable of operating successfully with
a hermetically sealed system. For this application it would probably be
best to undertake the development of the water~actuated hydraulic governor
system as the primary effort and the development of other types of gover=-
nor as the backup effort.

Even in an essentially constant~load plant, it is necessary to
regulate feedwater flow to the steam generator because of short-term
perturbations and longer term drifts in the system's behavior. Conven-
tional boiler water~level controllers and feedwater regulators contain
numerous sensors and control devices that would, if used, Jjeopardize the
operation of an unattended plant.

With a constant water inventory and a fairly steady load, feedwater
regulation can be obtained without the usual controllers by utilizing a
boiler feed pump that is quite sensitlve to suction head. The pump
capacity would vary in response to the depth of water accumulated in
the hot well. Performance curves for a submerged pump having such
characteristics are shown in Fig. 6. This pump is designed to regulate
the depth of liquid above the pump suction by the effect of suction head
on the degree of cavitation occurring at the impeller inlet. Pump
capacity is sharply reduced at low hot-well levels by partial vapori-
zation of the liquid at the impeller eye. With high suction heads in-
suring against cav1tation, the pump performance as total head vs capacity
is that represented by the dashed line 1n Fig. 6. A reduction of suction
head to below the level 1ndicated by the dashed line will initiate cavi-~
tation. Pump capacity will then be regulated by the suction head in

'accordance W1th the solid line.

Burns and Roe20 state that pumps of this design are currently in
use in ten large-capacity utility plants ' The pumps are designed to
operate with caV1tation and are expected to operate well in excess of a
year. ' '

' The use of a pump ‘that is sensitlve to the hot-well water level in

a flxed-lnventory system obviates the need for auxillary valves, level

49

 
 

 

 

Fig. 6.

4.6

UNCLASSIFIED
ORNL-LR-DWG 50111A

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

4.2

38
5, /)
x TOTAL HEAD VS PUMP CAPACITY /
+ 3.4 FOR NONCAVITATING FLOW /
- I
a /
<{
£
z /
b 3.0 300 f+ |
2
wn

/ CURVES OF PUMP CAPACITY VS
7 SUCTION HEAD FOR CONSTANT
/ TOTAL HEAD
2.6 7
//
2.2 350 f+
1.8
0 100 200 300 400 500

PUMP CAPACITY {gpm)

Feedwater Pump.

controls, sensors, or actuators.

system reliability is evident.

Characteristic Performance Curves for a Submerged Boiler

The importance of this to control

Other methods of achieving feedwater control and level control were

examined.

One scheme would employ overflow lines from a level drum in

the steam generator to the main condenser or to a feedwater heater.

Under normal operation the water level in the drum would remain between

two overflow outlets.

Thus some water could continuously recirculate to

the condenser through the lower overflow, while steam would recirculate

to the condenser through the upper overflow. A rise in drum level would

send water through the upper overflow and increase the recirculation,
while a fall in level would permit steam to exit at the lower overflow

and reduce the reclrculation,

system, or variations of it, could be made to keep the water level nearly

cbnstant.

50

By proper sizing of the return lines, this

The above system is inefficient in that water heated in the

Ab
 

 

 

 

"y

s

steam generator 1s recirculated back to the condenser without contributing
to the plant electrical output, but the loss might not be important in
the present gpplication.

Methods such as the use of eductors might also be possible for feed-
water flow control. The submerged pump, however, offers greater simplicity
and reliability than any other method considered.

Pumps. The conventional steam plant employs hot-«well condensate
pumps, boiler feed pumps, and sometimes forced-circulation pumps for the
steam generator. In the plant under study, consideration was given to a
system utilizing a single multistage pump to take care of all the feed-
water pumping functioas. The utilization of a single multistage pump
capable of operating from condenser pressure to steam generator pressure
has been demonstrated in the operation of SM=l. The additional modifi~
cation desired for this application is the development of a reliable
high-head low~flow canned=rotor pump to perform the required service.
The very successful operation of the canned-rotor primary pumps in
pressurized~water service indicates that the utilization of the canned-
rotor pump is completely compatible with high reliability requirements,
The attachment of a canned-rotor drive to a multistage impeller is a
combination which pump_manufacturers believe presents no difficulty or
extrapolation of known technology. If the required development were not
accomplished within the time available; a noncanned pump with the pump
leakage controlled and bled ‘back to the condenser could be used in a

"semiconventional system.
The poss1bility of incorporating the feedwater pump on the turbine

shaft and- thus eliminating a separate drive for the pump was considered.

This may represent a satisfactory method for pumping, but the departure
- from conventional de51gn might have a 51gnificant effect on the develop-
'~l;ment of the turbine-generator set and the pump,would have to operate at
_ ?the same speed and (if horizontal) at the same level as the turbine
:'seThen, too,ithe elimination of‘a high-reliability pump motor prdbably does
lVenot give any 51gnificant increase in over-all plant reliability

The emphasis placed on achieving utmost 51mp11City and reliability

for the reactor directed attention to the use of an injector rather than

51

 
 

 

 

a motor-driven feedwater pump. The relatively low temperature of con~
‘densate from the condenser (possibly 70 to 80°F) and the minor importance | (;;
of pumping efficiency are conditions well suited to the use of an injector.
It is particularly advantageous that injectors are of simple construction
and have ho moving parts. However, injectors have the disadvantages that
the drop in pumping efficiency at pressures above 300 psig necessitates
that large quantities of steam be used for pumping and that feedwater at
temperatures above 100 to 110°F cannot be handled by a single-stage system.
Present experience with injectors is such that they are not conSidered

to be entirely reliable for unattended operation, although this opinion

is in part based on the effect of load fluctuations on injector operation.
Aside from questions of operability and reliability, the feedwater regu-
lation obtained automatically with a cavitating pump mekes the pump more
atfiractifie than the injector. The inherent simplicity of the injector
suggests3.however, that it be studied further.

Heat Exchangers. The basic heat exchanger requirement for the

 

pressurized-water secondary system includes only the steam generator and
the main condenser. The use of feedwater heaters, regenerators, reheaters,
and similar equipment, which conventlionally are employed to increase the
plant thermal efficiency, would be largely contingent on their effect on
reliability. A few heat-recovery devices would not endanger system
reliability, but in general they should be kept to a minimum,

~ The vast experience in the construction of heat~exchange equipment
indicates that units of very high integrity can be fabricated and can
operate successfully without failure for periocds of more than one year.
As discussed in connection with the primary system, successful operation
depends heavily on a high level of quality control during fabrication,
but no extrapolation of current technology is required. The rather
modérate conditions of water temperature and pressure enviéioned for
the secondary system amplify the confidence that leaktight units capable
of one year of continuous operation can be constructed.

Potential materials of construction examined for heat exchanger

sérvice were the austenitic stainless steels, Inconel, and the copper- -

-‘nickel'alloys. Austenitic stainless steels can be used 1f one can insure

52
 

 

wj‘

&

operation of a system free of chlorides and/or maintain a very low level
of oxygen. The presence, however, of even small amounts of chlorides
and oxygen in such systems may result in crevice and stress-corrosion
cracking.?? On the other hand, many austenitic stainless steel systems
having acceptable control of water chemistry do exist and have operated
successfully for years without fallure of heat exchange equipment. In
this connection it should be noted that the SM=1l has operated since 1957
and has not encountered a single tube leak in any heat exchanger in the

system.11 The use of Inconel for steam generators in pressurized-water

- reactors has been investigated in recent work at Bettis.?” Inconel

apparently offers excellent resistance to both stress corrosion and
general corrosion. It is the preponderant\opinion of heat exchanger
manufacturers that materials can be selected and equipment can be fabri-
cated that can meet the one~yeaxr operational requirement.

In view of the preceding discussion, the proposed concept of a
hermetically sealed secondary system with a canned-rotor pump and a
turbine~generator as the only rotating machinery and a turbine-governor-
frequency control as the only control system appears promising for the
present application.

EFFECT OF ALTERED REQUIREMENTS ON THE DESIGN CONCEPT

The design concepts presented in this report are clearly a sensitive
function of the reqpirements for the nuclear power plant. The purpose
of this section is to point up those design areas that would require
careful re-examination to meet changes in the initial specifications.

TOf particular interest are the effects of the following assumptions:
;commnnication with the plant is possible, size and weight are considera=-
:fitions, unattended operation beyond one year 1s important the plant 1s
'iginot expendable, the time schedule is different; the load is variable.

Communication Wlth the Plant

If communication with the plant can be achieved, information perti-

' nent to plant operation-can be monitored. System.temperatures, electrical

53

 
 

 

 

trequency, and perhaps flow information could be transmitted to a receiving
center where elementary corrective actions could be initiated by simple
31gnals. Fixedsincrement adjustments of control rods would permit COr=
rection for long-term reactivity changes indicated by changes in water
- temperature. Similar1y3_fixed-increment ad justments of a gcvernor system
could correct for drift of the set point. Commmication also wowld permit
nonoperating spares to be remotely energized or emergency shutdown of the
plant to be initiated when received information indicated the need. The
hazard of misbperation and failure because of malfunctioning of the communi -
_cation,sysi:em wduld, of course, exist, but precautions, such as the use
of coincideht signals, might be employed to prevent spurlous actions.
Thus provision of a system for communicating with the plant might permit
Aimprovement of plant rellability.

| MMch of the broad simplification propbsed.to achleve relisbility |
for the unattended plant would be useful in desigping a plant‘for opera-
tion with small crews. Some of the modifications proposed for the control
end water-treatment systems might greatly reduce the operation and mainte=~

nance requirements of an attended plant.

Plant Size and Weight

 

Restriction of the size and weight probably would alter the design
of a reactor power plent in a menner that would adversely affect its
reliability. The design conslderations imposed by moderate restrictions
might have little effect, whereas overriding considerations of size and
weight would likely render reactors such as the pressurized-water concept
unsuitable. In designing a'preSSurized—water reactor for limited size
and welght, lncreased system pressure, temperatures, and flow veloclties,
higher reactor power density, and greater turbine-generator speeds would
have to be Investigated. 'Although incorporation of many of these'changés
would not necessarily decresse system reliebility, they might remove the
design from the realm where experlence with present pressurizeddwater'

'reactors is applicable.

54

-

O
 

 

 

)

Iong~Term Religbility and Expendebility

The design of an unattended power plant capable of operating several
years on a continuous baais imposes even more severe restrictions on the
systems and components. Particularly sensitive to longer term operation
would be the water-treatment requirements. Simple systems to permit
introduction of chemical additives might be needed, and side-siream
purification would probably be necessary.

The problem of bullding in and controlling sufficient reactivity
for longer term operation might necessitate the use of an automatlic
control system, although it is possible that the lifetime obtainable
using burnable poisons would extend well beyond one year. If automatic
control were required, 1t might he that & programmed shim-control system
operated by a clock or a watt«hour meter (as mentioned under Reactor
Control) would be more reliable than a system which employed feedback
from the reactor. Almost certainly, longer term reliebility would be
achieved if one proceeded to build plants initially capable of one year
of operation and then modified them as operating experience dictated.

Two major changes in concept would be required if the plant, or
even its major components, were not considered to be expendable. Equip-
ment such as the turbilneegenerator and the pumps would require protective
devices capable of shntting down the plant, and deposition of activity
might have to be controlled so that core components would be repairable.
These changes, particularly the use of protective devices, would decrease
the probability of the plant running & year without shutdown.

A compromise between expendability and repairability might be
attractive for some applications.' This could involve construction of
the plant in packages (say & core package, turbine-generator package,
etc ), which could e replaced as units. The cost might be much less

| than that of replacing an entire plant

Variable Loads |

The effect of reqniring an unattended.plant to operate successfully

.on & variable load might significantly change recommended components and

55

 
 

 

 

 

control concepts. The use of an induction generator, for instance, with
a variable load might produce voltage variations beyond the desired
limits. Similerly, certain frequency control systems (such as the
mechanical governor) might be less adaptable to a variable-load plant.
The effect of sudden load changes on reactor control requirements would
have to be examined. B

If the load were intermittent in nature, the requirements for the
plant would be even more stringent. Both the primary and the secofidary
systems would have to be capable of going to power unattended. Provision
of a variable dummy load to preclude zero-power operation and to'eliminate
severe load changes, if necessary, might improve the reliabllity of a
load=following system. Small or gradual load changes not requiring

.plant,shutd0wn could probably be accommodated by the design concepts

initislly proposed.

Effects of Time Schedule

 

Certainly one of the most fundamental restrictions affecting the
design concept is the time within which the»plant is required. This is
emphasized by the suggested effects of time changes on the concepts
presented. If production of asn acceptasble system in three years were
required, it is probable that success of the program would not be based
on design modifications beyond the semiconventional system proposed in
this report (although the hermetically sealed secondary system would be

attempted as a secondary effort). If, on the other hand, four to five

years were allowed for procurement of an acceptable plant, the hermetically

" sealed plent would certainly be utilized. For periods beyond five years,

one could afford to deviate to still greater degrees from known technolo-

gies and proven concepts.

SUMMARY AND CORCLUSIONS

Power reactors are normally designed for constant attention and

continuous maintensnce, and there are no existing reactors capable of

| producing electricity for one year unattended. Not only are'completé

56

 

 
 

 

 

 

-

®.y

oy

power plants incapable of one year of operation without maintenance, but
few of the individual components have the requisite reliability. This is
particularly true of control systems and of steam power systems, for
which regular maintenance is accepted practice. It 1s less true for the
components in the primary system of pressurized-water reactors, hecause
the Navy program has led to the development of reliable components.
Since existing systems do not have the required reliability, the
Teasibility of achieving an unattended reactor in four years actually
involves the question of whether a reliable system can be developed in
the allotted time. The four=year period is so short that there would be
little opportunity for perfection of new concepts, and the authors believe
a program to achieve the stated objectives should, wherever possible, be
based on proven concepts and proven technology. System rellability 1s
most likely to be achieved by simplification to the point that operation
depends on & minimum number of components. The objectives of this study
make simplification particularly promising, since the system is not re=
guired to be repairable, have a long life, follow & varying load, or
operate 1n a populous area. Although relisbility is not consistent with
minimum cost, elimination of_eqnipment by simplification may offset the

increased cost of individusl components.

A pressurized-water reactor fueled with highly~enriched uranium was
selected for this application because of the succegsful experience with
it and because 1t is amenable to extreme simplification. The selection
of the pressurizeddwater concept is not ‘intended to suggest that this is

the only or even.the best system for th;s service. It is, however, the

- one Wthh appears most likely of achievement in a limited time.

A reactor plant based on the concepts which evolved from this study

,_fi;lfwould consist of relatively few components. The only: mechanically
.--tfiloperating components would be the Primary coolant PumP: the boiler feed
{ffiV?P“sz and the turblnengenerator set, the turbine governor would be the
4: i only control SYStem. This simplificatlon'is obtained by m&king the

- 'ffollowing changes in an’ ordinary pressurlzed-water reactor plant:

l The reactor is designedrto operate after startup without a8

nuclear control systen. Burneble poison is used to limit gross reactivity

57

 
 

 

 

changes, and the negative temperature coefficient is used to cofipensate
for changes which are not eliminated. The effect of allowing the core
temperature to vary (say over s range of 50°F) is to shift the reactor
- control problem to the turbine governor. Since the governor is already
required for fréqpency regulation, no additional components are made
necessary.

The design of a turbine gbvernor and admlssion velve 1s eased by
elimination bf the normal requirement of tight frequency control over a
wide range of loads. Although seversl existing types of governor could
be modified for this system, the rellability of & governor is one of.the
'major system uncertainties. (The use of a variable dummy electric load
to control frequency without a thfottle valve or a governor appears to
| merit serious investigation.) .

- 2. An sutomatic control system for the pressurizer is avoided by a
-desig-fi-which-permits the electric heaters to remain on continuously;

3. DBoiler feedwater controls are eliminated by use of an existing
type of cavitating pump that is sensitive to suction head. The pump will
maintain a constant water level in the hot well and, consequently, in a
fixed-inventory system, will maintain a constant level in the steam
generator.

4, Both the primary and secondary systems are hermetically sealed,
and. there is neither water treatment nor blowdown. It appears feasible
to operate for a year in this manner without encountering disdbling
problems from corrosion or crud buildup.

The operation of a hermetically sealed secondary system requires
the use of water-lubricated turbine-generator bearings and a genersator
which will operate in a vapor or water environment. Several proven
brushless generators are capsble of such operation. While only a few
turbines in the slze range needed have been bullt with water=lubricated
bearings, canned~rotor pumps employing water bearings have been successfully

:’6pefatéd'for many years. (If development of water=-lubricated beafings -
'iWQQeS‘not.proceed rapldly enough, an oll-lubricated turbine with a brush-

 less generator can be accommodsted in a-modified'design.)

. 58 |

 
 

 

 

Wi

&3

&

The remaining components of the power plant appear to present no
rellability problems. Fuel elements with lifetimes longer than required
are available. Primary-circulating pumps have a demonstrated history of
relisbility. With careful design, selection of materials, fabrication,
and testing, the plping, heat exchangers, and pressurizer can be made to
be reliable.

It is probable thaf using the concepts proposed, reactors capable
of unattended operation can be available in four years. To achlieve this
objective, a conceptual design study must progress rapidly to the beginning
of detalled design, in order that the construction of a prototype can be
accomplished early in the program. Studies of water chemlstry, core
physics, and reactor control, and the development of a frequency=-control
system, tufbine-generatbr, and boiler feed pump should begin immediately.
A well-integrated and rapidly moving program can lead to useful power
plants in the allotted time. Experience with the first generation of
reactors and continuing development will increase the reliability beyond
what may be achieved initially.

59

 
 

 

 

 

10.

11.

12.

13.

15.

16.

17.

- 60

" REFERENCES

"Reliebility Study — Equipment and Components in Nuclear Power Plants s

'~ Gilvert Associates, Inc., Report No. 1512 (to be issued).

. - "APPR-1 Operating Experience," Nucleonics 17, 50-51 (1959).

M. L. Myers, "Study of Outage Experience with Selected Boiler' and
Turbine Generator Units," ORNL CF=60-3-56 (March 31, 1960).

"SL~1 Reactor Evaluat:.on, Final Report," p. V=22, IDO-19003 (July 19,
1959)

E. H. Bankcker et al., KAPL-2070 (March 7, 1960). (Cla.ssified)

"PM~1 Nuclear Power Plant Program, Third Quarterly Progress Report
1 September to 30 November 1959," p. III-133, MND-M-1814.

7. . Williamson, "APPR~1 Burnout Calculations, " APAE Memo 126
' ’(April 10, 1958).

"SMa-l Operating Contract Annual Report, February 1958—Fe'brua.ry 1959,
p. VII-23, APAE-47 (June 30, 1959).

A. Radkowsky, "Theory and Application of Burnable Poisons," 1958
Geneva Conference, Vol. 13, pp. 42645.

E. E. Gross, "Reactivity-Lifetime Comparison of U?3% and 10,000 MWD/Ton
Pu as Fuel for the APFR," ORNL CF-59-5-106 (May 21, 1959).

C. R. Feavyear, SM=l, personal commmnication, May 5, 1960.

D. J. DePaul, "Corrosion and Wear Handbook," pp. 33-36, TID-7006
(Maxrch 1957).

R. Hurst and S. McLain, "Technology and Engineering,” pp. 281—ff.,
Progress in Nuclear Energy Series, Pergamon Press, New York, 1956.

"The Shippingport Pressurized Water Reactor,” pp. 181—ff., Addison=-
Wesley Publishing Company, Reading, Massachusetts, 1958.

DePaul, op. cit., pp. 102-%.

DePaul, op. cit., pp. 229-38.

" G. P. Simon, "Van de Graff Study. Fourth Interim Report on the
‘Deposition of Corrosion Products Under Irradiation,"” pp. 134—46 in

WAPD-BT-16 (December 1959).

kb
 

 

 

 

 

 

 

o

%)

18.

19.

20.

2L.

22.

23.

24.

25.

26.

27.

28.

29.

G. P. Simon, "The Use of Neutron Activated Specimens for the Study
of Corrosion Product Release Rate,'" WAPD«T-1114 (March 1960).

B. G. Schultz, K. H. Vogel, and G. P. Simon, Bettis Plant, private
communication (June 1, 1960)

Burns and Roe, Inc., provided consulting service regarding certain
aspects of the secondary system. ‘

L. D. Damretowski, The Elliott Company, personal communication
(May 16, 1960).

J. G. Williams, Worthington Corporation, personal communication
(May 14, 1960).

R. E. Vula, J. A. Paget, and R. D. Winship, "Steam Generators for
Pressure Water Reactors," TDV-1 (AECL-832) (June 1959) , Section VII,
"Bibliographies on Boiler and Feedwater Technology."

Hurst and McLain, op. cit., p. 286.

S. Gordon and E., J. Hert, "Radiation Decomposition of Water Under
Static and Boiling Conditions," 1958 Geneve Conference Proceedings,
Vol. 29, pp. 13-18. | |

C. J. Hochanadel, "The Ré.a.i-ation Induced Reaction of Hydrogen and

~ Oxygen at 25°C—250°C," 1958 Geneva Conference Proceedings, Vol. 7,

pp. 521-5.

D. E. White, "Evaluation of Materials for Steam Generator Tubing,"
Pp. 156—96 in WAPD-BT«1l6 (December 1959).

Vuia, Paget, and Winship, op. cit., p. IV=4.

J. F. Wilkes, W. L. Denman, and M F Obrecht Amerlca.n Power Con=

 ference, Vol. XVIII, 1 527 (1955)

30,

._;‘DePaul, _2 cit. 5 pp. 188—-90.

61

 
 

 

 

 

. ORNL-2985
Reactors — Power
TID-4500 (15th ed.)

TINTERNAL DISTRIBUTION

1. L. G. Alexander 53. P. R. Kasten
2—11. C. F. Baes, Jr. , 54, C. P. Keim
12. R. J. Beaver 55.. M. T. Kelley
13. M. Bender o 56. R. B. Korsmeyer
l4. D. S. Billington 57. P. G. Lafyatis
15. E. P. Blizard 58. J. A. Lane
16. A. L. Boch 59. - R. S. Livingston
17. E. G. Bohlmann 60. M. I. Lundin
18. C. J. Borkowskl 6l. H. G. MacPherson
19. G. E. Boyd ' . 62. R. E. MacPherson
20. E. J. Breeding 63. W. D. Manly
21. R. B. Briggs 64. E. R. Mann -
22. C. E. Center (K-25) 65. H. C. McCurdy
23. R. A. Charpie ' '~ 66. H. F. McDuffie
24. H. C. Claiborne , 67. A. J. Miller
25. W. G. Cobb | 68. R. S. Miller
26. W. B. Cottrell 69. K. Z. Morgan :
27. J. A. Cox 70. J. P. Murray (Y¥-12)
28. F. L. Culler 71. M. L. Myers
29. H. N. Culver 72. F. H. Neill
30. H. E. Cunningham 73. M. L. Nelson
31. L. B. Emlet (K=25) 74. P. Patriarca
32. L. G. Epel 75. A. M. Perry
33. E. P. Epler 76. P. H. Pitkanen
34. M. H. Fontana 77. C. A. Preskitt
35. J. Foster 78. B. E. Prince
36. A. P. Fraas 79. M. E. Ramsey
37. J. H. Frye, Jx. 80—-89. R. C. Robertson
38. C. H. Gabbard 90—114. M. W. Rosenthal
39. J. L. Gabbard, ORGDP 115. G. Samuels
40. W. R. Gall 116. H. W. Savage
41, B. L. Greenstreet 117. A. W. Savolainen
42. J. C. Griess 118. L. D. Schaffer
43. W. R. Grimes 119. E. D. Shipley
44, E. E. Gross | 120. M, J. Skinner
45. E, Guth 121. S. H. Smiley (K-25)
46, R. J. Hefner 122. A. H. Snell
47. A, Hollaender 123. I. Spiewak
48. C. J. Hochanadel 124. F. J. Stanek
49, H. W. Hoffman 125, J. A. Swartout
50. A. S. Householder 126. W. C. Thurber
51. S. Jaye 127. M. Tobias
- 52. W. H. Jordan 128. D. B. Trauger
 

 

o

n

129.
130.
131.
132.
133.
134.

183.
184—185.

186.
187.
188.
189 *
190.
191.

192,
193.

194.

195.
196.
197.
198.
199.

200.
201,

- ..202.
L 203,
2044

J. J. Tudor 135—-144. M. M. Yarosh

E. Vincens . 145-176. Laboratory Records

C. 5. Walker 177. Laboratory Records, RC

A. M. Weinberg 1786~179. Central Research Library

G. D. Whitman 180—-182. ORNL — Y=12 Technical Library

C. E. Winters Document Reference Section

EXTERNAL DISTRIBUTION

W. B. Allred, Combustion Engineering, NRTS
'G. C. Banick, General Electric Company, 253 Main Street East,

Oak Ridge, Tennessee

A. E. Becker, Jr., Westinghouse Electric Corp., Lester Branch
P.0. Philadelphia 13, Pennsylvania

L. V. Bradnick, Pierce Governor Company, Anderson, Indlana
Gordon Brown, Alco Products, Dunkirk, New York

J. A. Brown, Allis Chalmers Mfg. Co., 702 Hamilton Bank
Building, Knoxville 2, Tennessee

R. L. Cawthon, Westinghouse Electric Corp., 605 Burwell
Building, Knoxville, Tennessee

M. A. Cordovi, International Nickel Company, 67 Wall Street,
New York 5, New York

F. Crim, GITTF, Fort Belvoir, Virginia

L. P. Damratowski, Elliott Company, Division of Carrier Corp.,
Jeannette, Pennsylvania .

R. 8. Etchison, Westinghouse: Electrlc Corp., Fast Pittsburgh,

Pennsylvania

C. R. Feavyear, SM«l, Fort Belvoir, Virginia

L. Flowers, Atomics International, Canoga Park, California
C. H. Fox, Martin Company, Baltimore, Maryland

J. G. Gallager, Alco Products, Schenectady, New York

M. Golik, Westinghouse Electric Corp., Bettis Plant

J. Gronan, Alco Products, -Schenectady, New York

V. V. Hebdrix, Ideho Operations Office
1F..Hittman, Martln Company, Baltimore, Maryland

C. A.. :Kennedy, Worthington Corp., Harrison, New. Jersey
L. A Kilgore, Westinghouse Electric Corp., East Pittsburgh,

lflj?ennsylvania

;”=f¥205;f@
7ff1“207'ji
' 5fBuilding, Pittsburgh,  Pennsylvania |
8. C." McGarry, Tdaho: Operations Office
‘R..L: Mela, Atomics International, Canoga Park, Californla o
_F. C. Miller, General Electric Company, Fitchburg, Massachusetts

;23—3;5-?; “F. Mollura, Rome Air Development Center, Rome, New York

T 208

Ji Ko Leslie, Alco Products, Schenectady, NeW'York

‘R. E. Lightle, Aerojet-General Nucleonics, NRTS
J. E. .McConnel, Westinghouse Electric Corp s Union Bank

63

 
 

 

 

 

212,
213.
214221,

222.
223.
224,
225.
226.
227.

228.
229.

230.

231.

- 232.
233.

234.
235.
236.
237.

238'

239.
240~816.

64

Ceptain R. L. Morgan, Idaho Operations Office

S. Neal, General Electric Company, Fitchburg, Massachusetts
Col. G. B. Page, Army Reactors Branch, Division of Reactor
Development, AEC, Washington

M. E. Peterson, General Electric Company, Schenectady,

New York
R. E. Ramsey, Burns & Roe, Inc., 160 W Broadway, New York 13,

New York

R. Rhudy, General Electric Company, Schenectady, New York
B. G. Schultz, Westinghouse Electric Corporation, Bettls Plant
V. D. Schwartz, Combustion Engineering, Windsor, Connecticut
B. Sellers, Elliott Company, Division of Carrier Corp.,
Jeannette, Pennsylvania

G. P. Simon, Westinghouse Electric Corporation, Bettis Plant
A. D. Sutton, General Electric Company, Fitchburg,
Massachusetts |

A. Swenson, Woodward Governor Company, Rockford, Illinois
K. H. Vogel, Westinghouse Electric Corporation, Bettis Plant

E. Wagman, Office of Naval Research, Washington

B. H. Webb, Westinghouse Electric Corporation, East Pittsburgh,
Pennsylvania

I. H. Welinsky, Westinghouse Electric Corporation, Bettis Plant
J. R. Wetch, Atomics International, Canoga Park, Californis

J. G. Williams, Worthington Corp., Newark, New Jersey

H. H. Wilson, Gilbert Associates, P.0. Box 1498, Reading,
Pennsylvania

P. C. Zmols, Combustion Engineering, Windsor, Connecticut
Division of Research and Development, AEC, ORO

Given distribution as shown in TID-4500 (15th ed.) under

~ Reactors — Power category (75 copies — OTS)

%

C