Nuclear Engineering and Design 156 (1995)

259 - 268

A review of the design and cons!ruclion of the Sizewell B

pressurized water reactor building containment stmctures
D.R. Davies a, AC. Roberts b, D. Ness b
aPWR Project Gr, up, Nuclear Elearic plc, Gk~~cesrer, UK
b NDA Consulring Engineers, E&mm!wn House, Tarton Sweet, Knuts/ard, Cheshire WA 16 6AF. UK

--

Abstract

The design and construction of the SizewellB primary containment has resuked in a structure dierent in certain
aspects from that envisa@ at the time when the decisionwas made for Sizewellto be based on the SNUPPS design
whichoriginatedin the USA.The successfulstructural overpressuretest (SOT)and integratedleak rate test (ILRT),
performedat the end of 1993, marked the completion of approximately 3 years of civil construction activity on the
containments. With the construction of the structure complete, it is a suitable time to review the d&gn of the
structure and to ascertain whether significant improvements can be gained, in terms of cost and construction
programme, for further projects based on the Skewelf design.

1. Mroduction presented. Referena is made to the effi of con-

struction conditions and programme on the de-
SizewellB is the first pressurized water reactor sign, and typical results from the structural
(PWR) nuclear power station to be built ia the overpressure test (SOT) are presented. Finally,
UK. Construction of the station for client Nu- areas where construction costs and programme
.__-. .
clearElectric(NE,) IS being managed from within could be eased for future projects are reviewed.
NE by the PWR Project Group; Nuclear Design
Associates (NDA) arc the design consultants re-
sponsible for the civil engineering works. The 2. Loading requirements and c~&&&xts
station design is based on the SNUPPS concept
with the primary Cloop circuit of Westinghouse The Sizewell B PWR primasy containment was
design with a grcss Dutput of 1260 MW. During designed using a UK design code whichwas based
the period of the design and construction of the on ASME (1983)with revisions up to and includ-
power station, there have been a number of ing the summer 1985Addenda. The load combi-
changes to the original design and to the form of nations in this code are generally similar to those
the final primary containment structure (Fig. 1). in ASME with the addition of a requirement to
Critical loading combinations are discussed and check the ultimate capacity of the containment to
details of two of the significant design changes withstand an internal pressure loading at ambient

0029~5493/95/$09.50
8 1995ElsevierScienceS.A. All rightsreserved
SSDI 0029-5493(94)00952-X
260 D.R. Davies et al. I Nucleur Engineering and Design 156 (1995j 259-268

/
1219m X.342 m RAU n166nlUD
I

Fig. I. Sectionthroughcontainmentstructures.

temperature to demonstrate an ultimate minimum Therefore, in response to an NII requirement to

safety factor of 2.0. clarify the design feasibility of the reactor building
During the preconstruction period, when the to withstand the combined effects of the design
design code was being developed, there was de- basis fault (DBF) pressure and SSE, an investiga-
bate (Stevenson, 1983) about the rationale of cou- tion was undertaken to determine the effect of
pling the internal design pressure loading with the coupling these load cases. This investigation
safe shutdown earthquake (SSE) load case, which showed that considering these load cases together
for Siiewell was specifiedas a maximum horizon- had significant implications on thti design of the
tal ground acceleration of 0.259, a maximum VU- containment but that the structure with the pravi-
tical acceleration of 2/3 the horizontal, and UK sion of additional reinforcement could accommo-
soft site spectra. The Nuclear Installations Inspec- date the coupled loading condition.
torate (NII), the UK licnsing authority, were :jf For this investigation and for follow-on design
the opinion that the two load cases should be purposes, the seismicrepresentation of the reactor
considered in combination, whereas the position building consisted of a beam and lumped mass
of the client, at that time called the Central Elec- model, A detailed examination of the results of
tricity Generating Board (CEGB), reflected the the seismic analysis showed that Exed base results
arguments for the decoupled case presented by were higher than the soil structure interaction
Stevenson’s paper (1984). (SSI) results, and that rocking of the containment
 D.R. Davies et al. I Nuclear Engineering and Design 1% (1995) 259-268 261

-0 I 2 3 4 .r 6
HEIGHT ABOVE BASE(m)

Fig. 2. Veritcal reinforcement required in the containmrnt wall.

structure due to the horizontal input dominated bined loading which required 3590 mm2 per metre
the response of the structure. This rocking was height at each face over the heightof the cylindri-
found to be sufficient to cause uplift under a cal barrel. In the lower part of the barrel. the
portion of the base, which for the combined load- maximumvertkal reinforcementat the wall base
ing ease made the response of the base to the junction was dictatedby the DBF combination
associated pressure loading more onerous. Design with a 1.5 load factor on design pressure P.
information was generated for the containment However,curtailmentof the reinforcementwas
wall using ;i pseudo-static analysis with a Fourier controlled by the combined DBF (1,OP)+ SSE
representation being used to apply the horizontal load combination(Fig. 2).
loading. The design of each section of the wall The ultimate load combination introdud 5.~
requiredthe combinedeffects of bending, axial the UK containment requires a demonstration
load and membraneshear to bc examined,and that the primary containment has the capacity to
additionalreinforcementrequirements wereestab- withstand a hypothetical internal pressure loading
lished. of at least twice the design pressure. This check
Finally an agreement was made between the was performed using a modified version of the
CEGB and the NII to design Sizewell B for the non-lirear program ADINA-81, the modiieations
combined loading condition. This decision was being made to improve the modelling of eonerete
made to enable the Sizewell design to proceed to behaviour. Similar modifications have since been
programme and the CEGB reserved their position incorporated in a later version of the program,
with respect to future stations. As predicted by ADINA-87. As part of the validation of the use
the earlier investigation, the combination of these of this programfor analysis of the Sizewell B
two load cases resulted in an increase in reinforee- containment,analyses were performedfor the
ment quantities throughout the containment. In NBC commissioned1/6th scale reinforcedcon-
the barrel,for example,the minimumhoop rein- crete containmentthat was built and tested at
forcementrequiredat each face, 2888mm2 per Sandia and for the 1110thscale prestressedcon-
metre height, was found to be sufficientfor all cretemodel of the Sizewelicontainmentthat was
load combinationswith the exceptionof the com- built and testedin the UK (Palfrey,1991).
262 D.R. Davies et al. / Nuclear Engineering and Design156(1995)259-268

Details of the final analyses and design of the

containment ilre curtained in the design report
which has beer)issuedto the NII. The 11volumes
of this report consist of a total of 55 individual
reports addressing specific items, or areas of the
containment. It provides a complete reviewof the
comprehensivedesign exercisecarried out on the
primary containment and includes supporting re-
ports suchas m the concretepropertytests,

3. Analysisof the primarycontainmentbuilding

Three computer codes were used for the finite cumLsvmtarMMB

element analysis of the primary containment: (i) Fig. 3. Pian view at SCEB support
cantilevers.
PAFEC for linear elast,icstatic analysis, (ii) NAS-
‘IRAN for seismic anaiysis, and (iii) ADINA for tainment that would otherwise be exposed tc the
non-linear analysis. environment. Together with the adjacent fuel and
Numerous models of various types were gener- auxiliary buildings it forms the secondary contain-
ated to investigate the general behaviour of the ment which has the function of providing an
structure, or the local effects at discontinuities, addiiionai barrier to any leakage originating from
load points or penetrations. Axisymmetricmodels the primary containment and its penetrations. In-
were generally used to examine the overall be- corporation of the SCEB was an enhancement to
haviour of the containment structure under the the original SNUPPS design.
full range of specifiedloading conditions. In addi- The secondary containment is formed of four
tion to stress output, for design of the contain- parts: (i) a iightweight concrete dome supported
ment, displacement and strain outputs were by reinforced concrete slabs (Figs. 3, 4 and 5)
obtained for use by the liner designer and by cantileveringout from the near top of the primary
various suppliers of equipment located within the containment barrel; (ii) a vertical reinforced con-
containment. crete wali over approximately a third of the cir-
The final analytical task for NDA on the
Sizewellproject was to re-analyst the containment
structure, to obtain the predicted strains and de-
formations for the structural overpressure test,
which was performed in December 1993. Unlike
the design analyses, these analyses included al-
lowances for the stiffening effects of the liner and
associated stiffeners, and the effects of reinforce-
ment and the prestressing tendons. Results from
long-term concrete tests were reviewed to produce
properties applicable to the age of the concrete
and the duration of the loading.

4. secoadsuy containment et&sure buIkIing

The secondary containment enclosure building Fig.4. View along arrow A (Fig. 3) showingarrangement of
(SCEB)envelopes the parts of the primary con- bearingsa ~1 prestressingbars.
 D. R. Davies et al. 1Nuclear Engineering and Design 156 (lPP5) 259-268 263

ment shell. Large gaps in the cantilever were also

provided at each of the three buttresses to allow
the permanent prestressing platforms to be used
to provide access to hoop tendon anchorages.
The design of the SCEB was governed by the
SSE load case. To minimix the seismiceffectsthe
dome was constructed with lightweight concrete
with a density of 1900kg rnd3. The loading on the
structure was determined by performing a NAS-
TRAN analysis usingthe modelin Fig.6. It was
found that the loading caused uplift on the bear-
ings between the cantilevers and the circumferm-
tial ringheam at the bottom of the dome. Vertical
prestressing bars, passing through the cantilevers
and into the ringbeam, were incorporated in the
design to provide an increase in the vertical reac-
tion on the hearings and to ensure that uplift was
prevented wnder seismicconditions.
The support conditions of the SCEB on the
cantilevers were examined in detail to arrive at the
most beneficial arrangements, with pinned, tixed
and partially fixed being considered with respect

Fig. 5. SectionB-B (Fig. 4) throughthe end of a cantilever.

cumference of the primary containment barrel;

(iii) reinforced concrete upstand walls on the roofs
of the adjacent buildings; (iv) enveloping build-
ings which are outside the structural description
provided in this paper.
It was originally proposed that the SCEB
would he constructed of a double layer of steel
panels, the inner being the pressure seal and the
outer providing weather protection and an archi-
tectural finish, supported off the surface of the
primary containment by a structural steel grillage.
However, the contractor offered a saving on the
tender price if the structure could be changed to
concrete and following a preliminary design exer-
cise the structure in its final form was shown to be
feasible and cost beneficial to the project.
lit was considered beneficial to keep the effects
of the change on the primary containment to a
minimum and hence the cantilever was divided
into a number of individual siabs in order to Fig. 6. Diagrammaticrepresentation
of the reactor building
reduce their hoop stiffening effect on the contain- seismicmodel.
 D.R. Davies et al. / Nuclear Engineering and Design 156 (1995) 259-268

to normal loadings and seismic response effects. the liner embedments increased and problems
The final result was the choice of a pinned ar- were anticipated with their fabrication and instal-
rangement utilizing prestressing bars and ‘pot’ lation at site. NDA proposed an alternative de-
bridge type bearings, the applied rotation between sign for a continuous reinforced concrete corbel
the top and bottom halves of each bearing being using through liner couplers to allow the reinforc-
accommodatedby a neoprene disk. ing bars to be continued into the containment
The wall forming the lower part of the SCEB is wall (Fig, 7). The corbel was divided into a total
restrained horizontally by a number of permanent of 48 individual units in order to reduce the local
radial props or ties spanning between the wall and increase in hoop stiffness in the containment wall,
the primary containment. It was divided into and the associated bending.
three vertical panels in order to reduce the hoop ‘Ibis change to the design required a number of
stiffness of the walls and the prop loads due to unconventional design features to be addressed.
radial movements of the wall and the primary Although the design of the corbel reinforcement
containment. Dividing the wall in this manner was based on established principles, the means of
also assisted construction on site as the outer anchoring the reinforcement through the liner and
panels could be constructed prior to closure of the the effect of lack of bond at the l&r-concrete
construction opening in the primary containment interface required investigation. The reinforce-
wall. Flexible pressure seals were installed in the ment was anchored using the through liner cou-
vertical gaps between the panels, and between the plers originally developedfor connections through
panels and the adjacent buildings. the containment bottom piate; however,appropri-
A further detailed NASTRAN model of the ate fatigue testing was carried out for the corbel
primary and secondary containments was pro- application.
ducedto establishthe finaldesign forces for the Details of the design and analysis of the corbels
seismic loads. This model included a representa- have been previously reported (Paton, 1991). In
tion of each individual support cantilever with addition, analyses were performed to determine
each bearing being representedby springs between the effectsof the corbels on the general behaviour
the cantilevers and the SCEB dome. Tbe loading of the containment. A mixture of atisymmetric,
on each component was therefore directly avail- Fourier and three-dimensional analyses was per-
able from the analysis. For design of the main formed to investigate a number of proposals to
reinforcement in the cantilevers, the radial load speed the construction of the corbels and the
and the bending moments about horizontal and installation and commissioning of the polar crane.
vertical axes all required consideration. The com- These analyses were complicated by the manner in
plete quadratic combination (CQC) method was which the concreting operations were being com-
used to combine the modal results rather than the pleted around the circumference of the wall, with
square root sum of squares (SRSS) method, the some segments being completed in advance of
SRSSmethod being unable to combine satisfacto- others.
rily the modes with close frequencies. The unique nature of the concrete corbel
prompted some concern from the licensing au-
thorities who requested confirmation that the de-
5. Polar crane corbels sign was adequate and that the liner and
containment were not jeopardized. Design reports
As per the SNUPPS design it was originally were submitted and the acceptable behaviour of
intended to support the polar crane rail on 32 the corbels was confirmed as predicted by com-
structural steel beams supported by steel liner paring theoretical movement and strain results
embedments. However, design changes were with readings taken during the commissioning
found necessary to the support arrangements ow- overload tests for the crane. Survey points were
iug to the thermal and seismic loadings on these located above the corbel so that movementscould
steel items. As a result the size and the weight of be directly measured during the test. In addition,
 D.R. Davies et al. / Nuclear Engineering and Design IS6 (i995) 259-268 265

i
4Dmn
T i

Fig. 7. Section through a polar crane eorbel.

strain readings were taken from instrumentation indicates acceleration of the barrel construction
embedded in the containment wall in the vicinity with increasing height. This progress was diet&d
of the corbel; this instrumentation was provided by the speed at which the reinforcementcould be
for the structural overpressure testing of the con- fixed in each pour. In the lower pours, where a
tainment. Both sets of readings confirmedthat the high density of reinforcement was required, it was
hr.l.n,,:,V.r
VCIiLiVI”b1
#Xrt.&-n,d..,J . ..%A-... ..A,” ,.mrn !.?,.AL.”
“1 b.,b b%tlirC‘ CuIUb, )iVuA‘ ilE,Li .“UUliiij rix0gr~iz.d hat fUfi&P &iays wklid OcCiUOwing
was acceptable. to liner works in complex areas preventing the
complete circumferenceof the wall being released
to the civil contractor. The reinforcement was
6. Constmetion programme detailed to allow flexibilityof working with some
sectors being fixed in advance of others. This
At an early stage in the design process the flexibilitywas extended to allow the contractor to
importance of the interfaces between the civil and pour concrete in a similar manner; the initial
liner contractors was recognized and steps were pours were inspected for signs of cracking near
taken to minimize potential problems. In the bar- the vertical construction joints and any cracks
rel it was recognized that the liner contractors’ found were monitored and were in fact only
work platforms could cause problems with the found to be thin and shallow. This uneven pro-
steel fixing operations below, and the decision was gress continued right to the top of the barrel after
taken to detail the reinforcement in lifts of 5.6 m which concreting returned to complete rings.
height. two pours, rather than the maximum bar Permanent shutters were detailed for the cor-
length of 12.0m. This eased reinforcement han- bels so that the need to erect falsework on the
dling and allowed the fixing operations to follow inside of the liner could be avoided, support being
as close to the liner works as possible. All joints provided by the concrete of the containment wall.
between reinforcing bars in the containment wall However, in the event, falsework was used to
and dome were made with CCL swaged couplers. altow concreting of the corbels to take place prior
The progress of the containment concreting to concreting of the associated containment wall
operations on site at Sizewell (Fig. 8) clearly pours. Work to the inside of the liner, to install
 D.R. Davies er al. / Nuclear Engineering and Design 156 (1995)259-268

Fig. 8. Progressof poursin the containmentbarrel.

the crane rail, could then proceed in parallel with tential problems with the proposed concreting
the work on the outside to complete the wall sequence.
pours. Considerab!e care was also taken with the
reinforcement detailing in the complex and con-
gested primary containment barrel at the height of 7. &uctura! overpressuretest
the SCEB support cantilevers and adjacent polar
crane corbels. This effort minimized the reinforce- The instrumentation for the SOT consisted of a
ment fixing problems and the programme date for mixkre of permanent instrumentation, cast into
the lift-in of the polar crane was successfully the concrete of the primary containment, and
achieved.’ temporary internal and external instrumentation
Owing to the programme requirement for set- which was installed only for the duration of the
ondaxy containment dome construction and the test, The permanent instrumentation consisted of
primary containment prestressing operations to vibrating wire strain gauges (VWSGs), foil strain
occur in parallel, the dome could not be concreted gauges, which were glued to lengths of reinforcing
in complete rings. Construction of the three dome bar prior to installation, stress gauges and ther-
areas ajdacent to the buttresses could only be mocouples. The temporary instrumentation used
started once the stressing operations on the dome mainly linear voltage displacement transducers
hoop tendons were completed, as temporary (LVDTs) measuring movements of invar bands
works to these areas would prevent access to the which were mounted between various points on
hoop prestressing anchorages. Therefore the three the inside and the outside of the building. Read-
parts of the dome above the cantilevers were ings from pressure gauges on the inside and the
azmstructedindependently in advance of the infill outside of the containment were used to deter-
panels, as was the top cap which was the first part mine the differential pressure that was being
of the dome to be constructed. Analyses were applied at all times during the test,
performed on the part completed dome in order During the test a monitoring system was in-
to determine built-in stresses and to identify po stalled which allowed all the gauge readings to be
 /
D.R. Duuies et 01. Nuclear Engineering and Design 156 (1995) 259-268 267

(MICROSTRUW

Fig. 9. Response of’ a typical hoop VWSG.

compared with their predicted values at the corre- 8. Design reviewfor future pmjects
sponding pressure. It was also possible to select a
gauge and to produce both screen and hard copy With the experience gained from the construe-
plots of the response from the gauge during the tion of Sizewell B, a number of areas can be
period from the start of the test. This continuous identified where it is cclnsidered possible to
mo.,itoring during the test enabled the time pe- achieve significant improvements in construction
riod of each of the pressure holds to be kept to a time and cost by making relatively minor changes
minimum. to the geometry of the structure. For the civil
Generally the behaviour of the primary con- work such savings have generally been identikd
tainment shell was within the predicted limits, as with reductions in the areas of the main retiorce-
can be seen by the typical results from a VWSG ment mats and with the reduction, or removal, of
shown in Fig. 9. The base was found to behave in the required shear reinforcement. The design re-
a stiffer manner than originally predicted by the view taking place includes an investigation into
analyses. It was considered that insufficient al- the provision of increased levels of prestressing
lowance had been taken of the stiffening effect of load and the corresponding reduction in rein-
the mass concrete that was placed before work forcement requirements.
started on the structural base. This was subse- The possibility of making reductions in the
quently confirmed by re-running the predictive reinforcement requirements is now being investi-
analyses with revised material properties for the gated in the following areas, which have been
elements in the area of the mass concrete. ideatified as being critical to speeding constmc-
268 DA Davies et ~1. / Nachr Enginewing and Design 156 (1995) 259-268

tion on future projects: (i) the base including the confidence to the ability of maintaining the con-
reactor cavity and instrumentation tunnel; (ii) the struction programme.
wall base junction; (iii) the equipment access pen-
etration; (iv) the SCEB cantilevers; (v) the polar
crane corbels. 9. conclusiou
As an example of reducing reinforcement com-
plexity, the base thickness is being examined in This paper provides a summary of the civil
order to establish its effect on the requirements in design features of the SizewellB containment. The
the base, reactor cavity and instrumentation tun- success of this probject is essential to the continm
nel, stressing gallery and wall base junction. The ued development of the nuclear industry in the
examination includes taking account of any :c- UK and it is ‘beiievedthat the Sizewell B design
ductions in the quantities of main reinforcement has provided a proven model on which to has
and shear reinforcement; it also takes account of further station designs.
any detailing changes that can be made to ease
the fixing of reinforcement on site.
The details of the polar crane corbel, and asso- Acknowledgement
ciated liner panels, are also being examined to
assess the effectof making the corbels continuous. The authors would like to express their thanks
This could result in a reduction in the size and to Nuclear Electric plc for their kind permission
number of the stiffeners on the liner, the rein- to produce this paper and to thank their col-
forcement in the corbel and the congestion in the leagues, both office and site based, for their assis-
containment wall caused by the radial reinforcing tance in its preparation.
bars. These savings must be balanced against
any increzscs ii; &G ~cinforcement required in the
wall caused by the additional bending due to Referems
the hoop stifbress of the corbel. Similar consider-
ASME Boiler and PressureVesselCode, Section III, Division
ation will be given to the structural form of the 2, Code for Concrete Reactor Vessels and Containments,
SCEB cantilevers which it is recognized require ASME, 1983, 51 edn.
optimization for potential changes in loading J. Palfrey and J.C.W. Smith, Ultimate load test of a l/lOth
criteria. scale model prestressed concrete containment, Trans,
Based on the experience gained with the con- SMIRT-I1, Tokyo, Vol. H, 1991, pp. 277-282.
A.A. Paton and A.K. Welch, Design and analysis of the
struction of the Sizewell B power station, NE is reinforced concrete support corbels for the polar crane in
optimistic that the construction programme for the reactor containment building of SizewellB power sta-
future contracts could be shortened from that tion, Civil Engineering in the Nuclear Industry. Thomas
adopted for SizewellB. Although the full effect of Telford, Windermere, 1991, pp. 187-200.
reducing the construction period for a particular J.D. Stevenson,Dcvelopu:entof concrete nuckar containment
(AC1 359) code requirements as they relate to probability
activity may not be fully gained on site, owing to based load combinations, Strxtural Engineering in Nuclear
other activities and interfaces becoming critical, Facilities,Vol. 2, American Society of Civil Engineers, 1984,
such savings should be pursued as they will add pp. 815-827.