Halden research reactor: 54 years young

Like France's Chooz A reactor, the Halden research reactor is built in a tunnel inside a cliff (Figures 1&2). Like other early nuclear power plants (it went critical in June 1959), it employs an unusual design: a natural-circulation boiling water reactor cooled with heavy water (deuterium). Maximum thermal power is 25 MW; peak temperature is 235°C and peak pressure is 34 Bar (3.4 MPa). The active height of its core is 80cm (2 feet). There are 300 (interchangeable) core positions for fuel and experiments, but to safeguard the integrity of the pressure vessel from radiation damage only the innermost 110 are used (Figure 3). There are typically about 30 experiments, and the rest of the positions are made up of driver assemblies, each with eight or nine UO2 fuel rods with 6% enrichment.

Performance data of materials held directly in the HWBR core are hardly applicable to the issues facing the world's fleet of light-water-based commercial reactors. So Halden staff have come up with a way of using its neutrons whilst mimicking PWR, BWR and CANDU reactor primary-circuit conditions. They have inserted special water circulation loops to pressure flasks mounted inside the reactor, into which instrumented test rigs are mounted. These loops employ equipment such as electric heaters and coolers, pressurizers, and purifiers (Figure 5). The Halden reactor currently has 11 such loops.

Test rigs consist of fuel or material samples attached to equipment such as bellows to create strain, gas lines to increase pressure, electric heaters or fuel rods for nuclear heating (Figure 4). Samples are attached to instruments such as thermocouples (measuring temperature and thermal conductivity), flow metres, pressure transducers (measuring fission gas release), cladding extensometers (axial stress), and linear voltage differential transformers (LVDT: converts movement into an electrical signal by using a magnet and electromagnetic field).

Unlike other research reactors, test assemblies can be monitored online as the experiments progress (their duration varies from a year to six years or more, for long-term tests of fresh fuel). More than 2000 signals from test rigs and the plant are logged at half-second intervals; raw data is backed up every minute, and converted data is entered into the test fuel databank every 15 minutes (5 minutes 2005-2012; 1 minute 2013+). Data from some 600 previous test experiments are available. Online data is supplemented by post-irradiation examinations (PIE) at the hot cells of the Kjeller headquarters of Norway's Institute for Energy Technology, the parent organization.

Critical for fuel and materials research

The accuracy of models and codes of fuel performance used for regulatory safety cases rely on these kinds of experimental data. For example, the UK's ENIGMA LWR fuel performance code is based on validation from Halden. Halden supplies about 23% of the overall amount of data supporting the UK National Nuclear Laboratory's fuel codes, according to Glyn Rossiter, senior research technologist, fuel cycle solutions, at NNL, and UK member of the Halden Programme Group. In some areas, such as fuel temperature measurements, it supplies 75-80% of data, he says. "It is easy to get data of the properties of materials not under irradiation, or to do post-irradiation examination afterwards. Halden is essential to get material property data of materials actually under irradiation."

The UK has been working with Halden since 1979. The lead signatory is now the National Nuclear Laboratory, and other members include EDF Energy, the regulator ONR, and vendor Rolls Royce. David Farrant of NNL, UK Halden lead and member of the Halden Board of Management, called Halden "the most important fuels and materials research lab for LWR fuel anywhere." He adds that although the UK has signed up to support the Jules Horowitz Reactor currently under construction in Cadarache, France, involvement in that project is no reason for the UK research community to turn its back on Halden. Given the length of time required to develop a pedigree of reliable data, "we will need to be members of Halden for at least 10 to 20 years," he says.

The Halden Reactor Project's reactor and facilities are affiliated to the OECD's Nuclear Energy Agency, but they are hosted and operated by IFE, which employs 270 people. HRP is about 50% of IFE's activities in Halden, Norway. Its annual operational budget is about EUR 17 million; funding comes by subscription from 14 international signatory country members, and 18 associated party members from 10 countries (they comprise in total more than 100 nuclear organizations worldwide). Supporters are mostly European, plus Japan, Korea, Russia, Kazakhstan, the UAE and the USA. These members fund collaborative research projects, whose results are available to all parties, and so-called 'bilateral' research projects whose results are private. This latter group provides essential financial support for the project: although Norway does not use commercial nuclear power, it hosts the reactor and the government stumps up 30% of the total HRP budget.

The next three years

The reactor's current three-year experimental programme runs through the end of 2014; its plans for 2015-17 were launched earlier this year and are now being pitched to members. Research priorities will be ranked and a project plan will be finalized in June 2014, for official signing in December 2014. The proposed programme totals 30 unique tests: 22 in fuels, eight in materials. However since the expected budget will not cover all of these (there are 23 in the 2012-2014 programme), the least popular will have to be dropped.

In addition, the reactor needs a new 10-year operating licence from 2015. After two years' work, an application was filed in December 2012. Although the process is not completed, there have been 'no negative signals' from the Norwegian regulator so far, according to HRP general manager Fridjov Øwre. In October 2013 the Norwegian government gave an in-principle decision to carry on with HRP, given sufficient international interest in the project. This month, the Halden management board will state its intention to carry on.

Research at Halden is divided into essentially two strands: in-pile fuel and materials testing in the reactor, and man-technology-organization (MTO) human factors research in experimental facilities, which include a reactor operator testing area, the Halden man-machine laboratory (Hammlab), a virtual reality centre and technology test centre (FutureLab). The 2015-17 MTO research programme includes expanded research on human performance in emergency and accident situations, and an emphasis on outage management and decommissioning. It is also looking at software systems dependability, including software development, assurance and certification.

Fuel and materials testing research

Fuel and materials testing research falls into several categories: fuel safety, cladding performance and plant ageing.
Fuel safety is examined in different ways: as integral effects (taking into account all of the complex and interrelated phenomena), in separate effects studies that aim to isolate a particular phenomenon, fuel behaviour in transients, and in cladding performance.

Integral effects tests

Integral fuel performance includes long-term monitoring of VVER fuel performance (pellet-cladding mechanical interaction, thermal behaviour and fission gas release) from fresh, including a doped rod with 5% by weight Gd, and a large-grain fuel version, up to a burnup of 60 MWd/kgUO2. At that point, these rods will be moved to a higher-power area of the Halden core to further study fission-gas release up to 65 MWd/kgUO2.

Another integral test, project IFA-716, manipulates reactor power to examine the fission gas release threshold, since it is a performance-limiting factor. Six fuel rods with different grain sizes and dopants (including a reference rod with 10 µm standard grain size and four rods with large (50-70 µm) grains created by adding chromium oxide or special sintering) are irradiated at or near the threshold. In these fuels, the effect of large diffusion length from the large grains competes with the increased diffusion coefficients from the dopants. One other fuel type included is UO2 in a beryllium oxide matrix, whose greater thermal conductivity is expected to reduce fuel temperatures and fission gas release. And in fact early results (see Figure 6) seem to indicate just that.

A new long-term fuel integral irradiation experiment is due to start in 2015 from fresh fuel; candidate fuel materials could be uranium nitride, uranium fuel with beryllium oxide whiskers, uranium fuel with liquid metal in the fuel-clad gap, fuel with radial variation in grain size, and fuels with varying levels of additives or levels of enrichment.

High-burn-up fuel refabricated from actual commercial NPPs is studied in the IFA-720 experiment. Its integral performance (again in terms of thermal properties, fission gas release and pellet-clad mechanical interaction) is studied during steady-state conditions and power ramps. This test rig incorporates a tubular coil around its external edge that can be pressurized with helium-3 (a strong neutron absorber) to control the power in the test and so provide greater precision of the required experimental conditions. The next test currently being prepared will use ~48 MWd/kgUO2 burn-up fuel with 8% by weight Gd.

Separate effect tests

A significant separate effect test aims to establish maximum tolerable rod overpressure without exceeding safety criteria limits. In the test rig, one high-burnup (50-70 MWd/kgUO2) fuel rod (PWR, BWR or VVER) is placed in the centre of a pressure vessel surrounded by booster fuel rods to provide local fast flux. Temperature is measured via a fuel thermocouple drilled into the rod, whilst a gas line provides pressures up to 600 bar (350 bar beyond PWR levels). Instruments measure clad lift-off and cladding elongation (to determine the degree of PCMI). Results have indicated that fuel rods can tolerate overpressures greater than 100 bar before cladding lift-off occurs. Current plans for this project include testing cladding types with different resistance to creep (permanent deformations): one recrystallised, one stress-relief annealed.

Another separate effects test aims to identify the amount of creep (slow permanent deformation) in fuel, to help better understand PCMI. Experiments have shown that irradiated UO2 and MOX fuel creep depends on applied stress and fission rate, but does not depend on temperature up to 1000°C. A current test monitors creep not in fuel pellets, but in thin (1.7mm) slices of LEU fuel sandwiched between 3mm-thick molybdenum discs. This set-up improves heat transfer to create nearly isothermal conditions in the discs. A load can be applied to the fuel stack with a bellows. The testing protocol varies temperature and applied pressure to yield data on fuel elongation over time (it also measures fuel densification and swelling). The current test uses chromium-doped fuel; future test plans include using gadolinium-doped fuel, MOX, and pushing to higher temperatures.

Halden tests on high-burnup fuel (above a threshold of about 60 MWd/kg) have shown a tendency for the fuel to fracture into small pieces (tens to hundreds of microns) which relocate within the fuel rod in LOCA simulations. Tests have also shown that this fracturing does not happen if a tight contact remains between cladding and fuel pellet. A test carried out in October 2013 yielded interesting results. In the standard setup, a single high-burnup fuel rod is placed inside a pressure flask; decay heating is simulated with a low level of nuclear heating. A transient is simulated; after a blowdown phase, fuel rod pressure rises until failure. A gamma scan of the fuel rods showed that the cladding swells like a balloon, and that small pieces of fractured fuel can be seen to have moved into the ballooned area. When the rod surface bursts open, pieces of fractured fuel are expelled. The research question here is, does the shock of the tube failure fracture the fuel, or is it fractured already? Future tests will slow the ramp-up to try to balloon the fuel rod but stop the test before it ruptures.

After long periods of irradiation, the edges of UO2 fuel pellets have been shown to form a high-burnup structure (HBS), generating plutonium and a local power and burnup peak. A test in conditions of fast increase in power and temperature (power transients) will study fission gas release in HBS fuel irradiated in a previous programme. Two UO2 rods with a burnup of 120 MWd/kg have been subjected to these transients, and two more 200 MWd/kg burnup rods completing an irradiation run would soon be available for refabrication and testing, according to a Halden planning document.

Fuel cladding corrosion and hydriding is also being investigated. Longer operating cycles and greater fuel burn-ups put pressure on cooling water chemistry; greater amounts of boron are required for reactivity control, and therefore greater amounts of lithium are required to achieve the required pH. However, high levels of lithium (above 3.5 ppm) have generally not been implemented in commercial power reactors; aggressive corrosion could possibly penetrate the fuel barrier. The experiment would raise lithium concentrations to 10ppm to look for cliff-edge effects in the corrosion mechanism.

A proposed research project would aim to investigate the stress on spent fuel caused by interim dry storage. A lot of fuel is expected to go from wet to dry storage without final disposal for 100 years (up to 300); utilities need to be certain that whatever fuel they put in storage will remain in the same condition. However, fuel assemblies may experience high temperatures (400°C), and high cladding hoop stress in storage. It is possible that there could be a cladding failure due to creep rupture or delayed hydride cracking. Halden proposes experiments in a simulated spent fuel canister, and performing accelerated testing. However, the testing regimen still needs to be defined.

Another experiment due to start in 2014 (IFA-772) measures crack growth rates in irradiated specimens of cold-worked 316, 321 and 304 stainless steel taken from actual power plant materials (such as control blades, baffle bolts, the core shroud). These specimens are machined to create a notch, equipped with a bellows for variable load tension, and instrumented according to the reversing DC potential drop method. In this method, a lengthening crack changes the signal; through calibration work researchers have been able to pinpoint the crack growth rate. This work could be used to study the impact of changes in water chemistry, for example hydrogen water chemistry in BWRs, on cracking behaviour. Other material testing programmes at Halden include crack initiation studies and pressure vessel integrity studies using punch tests.

Will Dalrymple is editor of Nuclear Engineering International magazine.