Thesis (MScEng (Mechanical and Mechatronic Engineering))--University of Stellenbosch, 2010. / ENGLISH ABSTRACT: A reactor cavity cooling system (RCCS) is used in the PBMR to protect the concrete citadel surrounding the reactor from direct nuclear radiation impingement and heat. The speci ed maximum operating temperature of the concrete structure is 65 ±C for normal operating conditions and 125 ±C for emergency shut-down conditions. A conceptual design of an entirely passive RCCS suitable for the PBMR was done by using closed loop thermosyphon heat pipes (CLTHPs) to remove heat from a radiation heat shield over a horizontal distance to an annular cooling dam placed around the PBMR. The radiation shield is placed in the air space between the Reactor Pressure Vessel (RPV) and the concrete citadel, 180 mm from the concrete citadel. A theoretical heat transfer model of the RCCS was created. The theoretical model was used to develop a computer program to simulate the transient RCCS response during normal reactor operation, when the RCCS must remove the excess generated heat from the reactor cavity and during emergency shut-down conditions, when the RCCS must remove the decay heat from the reactor cavity. The main purpose of the theoretical model is to predict the surface temperature of the concrete citadel for di erent heat generation modes in the reactor core and ambient conditions. The theoretical model assumes a 1D geometry of the RCCS. Heat transfer by both radiation and convection from the RPV to the radiation heat shield (HS) is calculated. The heat shield is modelled as a n. The n e ciency was determined with the experimental work. Conduction through the n is considered in the horizontal direction only. The concrete structure surface is heated by radiation from the outer surface of the heat shield as well as by convection heat transfer from the air between the heat shield and the concrete structure surface. The modelling of the natural convection closed loop thermosyphon heat pipes in the RCCS is done by using the Boussinesq approximation and the homogeneous ow model. An experiment was built to verify the theoretical model. The experiment is a full scale model of the PBMR in the horizontal, or main heat transfer, direction, but is only a 2 m high section. The experiments showed that the convection heat transfer between the RPV and the HS cannot be modelled with simple natural convection theory. A Nusselt number correlation developed especially for natural convection in enclosed rectangles found in literature was used to model the convection heat transfer. The Nusselt number was approximately 3 times higher than that which classic convection theory suggested. An optimisation procedure was developed where 121 di erent combinations of n sizes and heat pipe sizes could be used to construct a RCCS once a cooling dam size was chosen. The purpose of the optimisation was to nd the RCCS with the lowest total mass. A cooling dam with a diameter of 50 m was chosen. The optimal RCCS radiation heat shield that operates with the working uid only in single phase has 243 closed loop thermosyphon heat pipes constructed from 62.72 mm ID pipes and 25 mm wide atbar ns. The total mass of the single phase RCCS is 225 tons. The maximum concrete structure temperature is 62.5 ±C under normal operating conditions, 65.8 ±C during a PLOFC emergency shut-down condition and 80.9 ±C during a DLOFC emergency shut-down condition. In the case where one CLTHP fails and the adjacent two must compensate for the loss of cooling capacity, the maximum concrete structure temperature for a DLOFC emergency shut-down will be 87.4 ±C. This is 37.6 ±C below the speci ed maximum temperature of 125 ±C. The RCCS design is further improved when boiling of the working uid is induced in the CLTHP. The optimal RCCS radiation heat shield that operates with the working uid in a liquid-vapour mixture, or two phase ow, has 338 closed loop thermosyphon heat pipes constructed from 38.1 mm ID pipes and 20 mm wide atbar ns. The total mass of the two phase RCCS is 198 tons, 27 tons less than the single phase RCCS. The maximum concrete structure temperature is 60 ±C under normal operating conditions, 2.5 ±C below that of the single phase RCCS. During a PLOFC emergency shut-down condition, the maximum concrete structure temperature is 62.3 ±C, 3.5 ±C below that of the single phase RCCS and still below the normal operating temperature of the single phase RCCS. By inducing two phase ow in the CLTHP, the maximum temperature of the working uid is xed equal to the saturation temperature of the working uid at the vacuum pressure. This property of water is used to limit the concrete structure temperature. This e ect is seen in the transient response of the RCCS where the concrete structure temperature increases until boiling of the working uid starts and then the concrete structure temperature becomes constant irrespective of the heat load on the RCCS. An increased heat load increases the quality of the working uid liquid-vapour mixture. Working uid qualities approaching unity causes numerical instabilities in the theoretical model. The theoretical model cannot capture the heat transfer to a control volume with a density lower than approximately 20 kg/m3. This limits the extent to which the two phase RCCS can be optimised. Recommendations are made relating to future work on how to improve the theoretical model in particular the convection modelling in the reactor cavities as well as the two phase ow of the working uid. Further recommendations are made on how to improve the basic design of the heat shield as well as the cooling section of the CLTHPs. / AFRIKAANSE OPSOMMING: 'n Reaktor lug spasie verkoelingstelsel (RLSVS) word in die PBMR gebruik om die beton wat die reaktor omring te beskerm teen direkte stralingskade en hitte. Die gespesi seerde maksimum temperatuur van die beton is 65 ±C onder normale bedryfstoestande en 125 ±C gedurende die noodtoestand afskakeling van die reaktor. 'n Konseptuele ontwerp van 'n geheel en al passiewe RLSVS geskik vir die PBMR is gedoen deur gebruik te maak van geslote lus termo-sifon (GLTSe) om hitte van die stralingskerm te verwyder oor a horisontale afstand na 'n ringvormige verkoelingsdam wat rondom die reaktor geposisioneer is. Die stralingskerm word in die lug spasie tussen die reaktor drukvat (RDV) en die beton geplaas, 180 mm vanaf die beton. 'n Teoretiese hitteoordrag model van die RLSVS was geskep. Die teoretiese model was gebruik vir die ontwikkeling van 'n rekenaar program wat die transiënte gedrag van die RLSVS sal simuleer gedurende normale bedryfstoestande, waar die oorskot gegenereerde hitte verwyder moet word vanuit die reaktor lug spasie, asook gedurende noodtoestand afskakeling van die reaktor, waar die afnemingshitte verwyder moet word. Die primêre doel van die teoretiese model is om the oppervlak temperatuur van die beton te voorspel onder verskillende bedryfstoestande asook verskillende omgewingstoestande. Die teoretiese model aanvaar 'n 1D geometrie van die RLSVS. Hitte oordrag d.m.v. straling asook konveksie vanaf die RDV na die stralingskerm word bereken. The stralingskerm word gemodelleer as 'n vin. Die vin doeltre endheid was bepaal met die eksperimente wat gedoen was. Hitte geleiding in die vin was slegs bereken in die horisontale rigting. Die beton word verhit deur straling vanaf die agterkant van die stralingskerm asook deur konveksie vanaf die lug tussen die stralingskerm en die beton. The modellering van die natuurlike konveksie GLTS hitte pype word gedoen deur om gebruik te maak van die Boussinesq benadering en die homogene vloei model. 'n Eksperiment was vervaardig om the teoretiese model te veri eer. Die eksperiment is 'n volskaal model van die PBMR in die horisontale, of hoof hitteoordrag, rigting, maar is net 'n 2 m hoë snit. Die eksperimente het gewys dat die konveksie hitte oordrag tussen die RDV en die stralingskerm nie met gewone konveksie teorie gemodelleer kan word nie. 'n Nusselt getal uitdrukking wat spesi ek ontwikkel is vir natuurlike konveksie in geslote, reghoekige luggapings wat in die literatuur gevind was, was gebruik om die konveksie hitteoordrag te modelleer. Die Nusselt getal was ongeveer 3 maal groter as wat klassieke konveksie teorie voorspel het. 'n Optimeringsprosedure was ontwikkel waar 121 verskillende kombinasies van vin breedtes en pyp groottes wat gebruik kan word om 'n RLSVS te vervaardig nadat 'n toepaslike verkoelingsdam diameter gekies is. Die doel van die optimering was om die RLSVS te ontwerp wat die laagste totale massa het. 'n Verkoelingsdam diameter van 50 m was gekies. Die optimale RLSVS stralingskerm, waarvan die vloeier slegs in die vloeistof fase bly, bestaan uit 243 GLTSe wat van 62.72 mm binne diameter pype vervaardig is met 25 mm breë vinne. The totale massa van die enkel fase RLSVS is 225 ton. Die maksimum beton temperatuur is 62.5 ±C vir normale bedryfstoestande, 65.8 ±C vir 'n PLOFC noodtoestand afskakeling en is 80.9 ±C vir 'n DLOFC noodtoestand afskakeling. In die geval waar een GLTS faal gedurende 'n DLOFC noodtoestand afskakeling en die twee naasgeleë GLTSe moet kompenseer vir die vermindering in verkoelings kapasiteit, is die maksimum beton temperatuur 87.4 ±C. Dit is 37.6 ±C laer as die gespesi seerde maksimum temperatuur van 125 ±C. Die RLSVS ontwerp kan verder verbeter word wanneer die vloeier in die GLTSe kook. Die optimale RLSVS stralingskerm met die vloeier wat kook, of in twee fase vloei is, bestaan uit 338 GLTSe wat van 38.1 mm binne diameter pype vervaardig is met 20 mm breë vinne. The totale massa van die twee fase vloei RLSVS is 198 ton, 27 ton ligter as die enkel fase RLSVS. Die maksimum beton temperatuur is 60 ±C vir normale bedryfstoestande, 2.5 ±C laer as die enkel fase RLSVS. Gedurende 'n PLOFC noodtoestand afskakeling is die maksimum beton temperatuur 62.3 ±C, 3.5 ±C laer as die enkel fase RLSVS en nogtans onder die maksimum beton temperatuur van die enkel fase RLSVS vir normale bedryfstoestande. Deur om koking te veroorsaak in die GLTS word die maksimum temperatuur van die vloeier vasgepen gelyk aan die versadigings temperatuur van die vloeier by die vakuüm druk. Hierdie einskap van water word gebruik om 'n limiet te sit op die maksimum temperatuur van die beton. Hierdie e ek kan gesien word in die transiënte gedrag van die RLSVS waar die beton temperatuur styg tot en met koking plaasvind en dan konstant raak ongeag van die hitte belasting op die RLSVS. 'n Toename in die hitte belasting veroorsaak net 'n toename in die kwaliteit van die vloeistof-gas mengsel. Mengsel kwaliteite van 1 nader veroorsaak numeriese onstabiliteite in die teoretiese model. The teoretiese model kan nie die hitteoordrag beskryf na 'n kontrole volume wat 'n digtheid het laer as ongeveer 20 kg/m3. Hierdie plaas 'n limiet op die optimering van die twee fase RLSVS. Aanbevelings was gemaak met betrekking tot toekomstige werk aangaande die verbetering van die teoretiese model met spesi eke klem op die modellering van konveksie in die reaktor asook die modellering van twee fase vloei. Verdere aanbevelings was gemaak aangaande die verbetering van die stralingskerm ontwerp asook die ontwerp van die verkoeling van die GLTSe.