Eyeing PCI on the fly

Methods and codes for calculating the thermal-mechanical performance of fuel rods under irradiation, and for establishing manoeuvring guidelines to assure that the fuel does not exceed design limits, have been evolving for many years. They vary in complexity from high-order thermal-mechanical codes that are able to track a large number of parameters acting on a detailed geometry, to rule-based algorithms. Historically, power-manoeuvring tools for avoiding pellet cladding interaction (PCI) failures have consisted of rule-based algorithms that are well-rooted in experience. However, their success depends on updates to incorporate lessons learned from recent fuel failures, and they require conservatism to compensate for their simplicity.

Areva has recently developed a new type of power manoeuvring software which works in conjunction with the normal core monitoring code to calculate the stresses imposed on fuel rod cladding by a uranium pellet expanding with increasing temperature.

Tennessee Valley Authority has completed beta testing of this software, XEDOR from Areva, at its Browns Ferry boiling water reactor-based nuclear power plant in Alabama.

“For the first time, this new technology allows the reactor engineers to plan and execute reactor power manoeuvres with an explicit understanding of the stresses imposed on the fuel rod cladding,” says Jim Lemons, TVA senior manager, reactor engineering and fuels.

Browns Ferry has been using XEDOR software simultaneously with the current rule-based power manoeuvring protection, and has benefited from the new system.

“We have found that we are able to take advantage of the excess conservatism that was inherent in the current method,” he says.

Stress calculations

XEDOR is based on Areva’s high-order code RODEX4, which has been approved by the US Nuclear Regulatory Commission. RODEX4 models an entire fuel rod with high geometric resolution and multi-dimensional effects. However, it requires computer resources that are prohibitively large for online applications.

XEDOR calculates the stress through the use of a reduced-order fuel rod thermal-mechanical model, which tracks a smaller number of parameters to economise on computer resources. For example, unlike RODEX4, XEDOR in its current form is not set up to calculate internal gas pressure or fission gas release from the irradiated fuel pellets. Also, XEDOR applies a one- dimensional stress-strain model to calculate cladding hoop stress, while RODEX4 calculates all stress components. Still, Areva claims that its calculation of cladding hoop stress, the key PCI parameter, has comparable accuracy. (XEDOR stress calculation results were also found to be comparable to the high-order FALCON code by Anatech as part of the Zero Failure By 2010 programme.)

The current method of monitoring fuel for PCI is essentially rule-based. These rules allow local power levels in the fuel bundles to reach a certain level before having to condition the fuel by controlling the rate of local power increase. The rules are derived from calculations using higher order thermal-mechanical code. However, the evaluations are generic in nature, and apply to all reactors in all situations. The rule-based monitoring does not look at every individual fuel rod segment in a node, but only looks at what is predicted to be the peak power fuel rod segment, and the rod predicted to be most affected by pulling a control rod adjacent to the assembly.

“While the current rule-based approach has been quite successful in reducing the incidence of PCI failures for fuel manufactured by Areva, it can in some cases be overly conservative. For example, certain manoeuvring rules related to the allowable rate of power increase do not factor in the local hoop stress of the fuel,” Areva says.

XEDOR improves the rule-based approach by calculating the stress in every fuel node within every fuel rod in the core, based on the actual conditions in the core at that time. Also, Areva says that unlike the rule-based approach, XEDOR is never non-conservative.

One of the principal goals of the XEDOR testing programme carried out at Browns Ferry was to compare the XEDOR approach to the rule-based monitoring approach. Manoeuvres still have to comply with rule-based conditioning (which is currently the Areva fuel warranty basis).

In general, XEDOR confirmed that the restrictions imposed by the rule-based approach are conservative, and some relaxation of power ramping requirements is possible in most cases. However, the testing shows that there are cases where the rule-based approach is not quite as conservative as desired. These observations have been part of the feedback provided to the vendor.

“While XEDOR is not yet the official fuel-manoeuvring tool for Areva, TVA has realized some benefits during this testing phase. For example, TVA has imposed certain additional administrative restrictions (not required by the rule-based monitoring) on our fuel preconditioning. XEDOR has been used to assess these administrative restrictions, and in some cases we have been able to relax or eliminate them,” says Greg Storey, manager of TVA boiling water reactor fuel engineering. “In general, we have been able to justify not taking a small power drop for control rod exercise testing. For example, in a recent beginning of cycle startup, we were able to eliminate an entire load drop and power ramp to set the final control rod pattern.”

The software can reduce the depth of load drops and allow faster power ramps, so it can produce an increase in capacity factor, while ensuring fuel integrity by monitoring the core with a higher degree of rigour, Areva says.

Using the software

XEDOR is integrated into the Areva core monitoring software (POWERPLEX-XD), so it runs on the same platform. The software works in conjunction with the core monitoring code to calculate the power distribution in the core, down to the level of individual fuel nodes. (Note that for calculation purposes, a typical BWR active fuel rod, measuring 150 inches (381 mm), is divided into 25 nodes or segments. There are approximately a dozen pellets represented by each 6-inch node.) The detailed power distribution is supplied to XEDOR, which then calculates the clad stress for each fuel rod segment within each node.

The software can be used in real time to predict stress levels in current operation (online, as at TVA), or offline, to predict stress levels in a hypothetical case, or to evaluate fuel failures as a forensic tool. While TVA remains the only online case, many other stations are using the software offline (see figure for example).

If XEDOR is being used in active core monitoring, then it is fed real time information from the core monitoring code. When the operator requests a core monitoring case, the core monitoring software (POWERPLEX-XD) runs the 3D core simulator code (MICROBURN-B2) to calculate the detailed core power distribution. MICROBURN-B2 draws on plant information such as feedwater flow, feedwater temperature, dome pressure, control rod positions and jet pump drive flow along with local power range monitor (LPRM) neutron flux instrumentation. Once the 3D simulator code has calculated the power distribution, it then passes the results to XEDOR, which performs the detailed cladding stress calculation.

If a typical core has 764 fuel bundles, each with 91 rods, and each rod has 25 segments, a time step computation requires 764x91x25=1,738,100 calculations. Computation time varies depending on the computer hardware installed in the plant; typically a few seconds are sufficient to perform a time step computation.

The output of XEDOR shows the user how close the limiting fuel nodes are to the clad stress limits, as well as providing information on the relative stress distribution in the core. XEDOR is applicable to all BWR fuel types and can be used with other vendors’ software.

Going forward

TVA plans to make XEDOR its official fuel conditioning monitor method, and the official fuel warranty basis at Browns Ferry. Because Areva recommends implementing XEDOR at the start of a new cycle, TVA has scheduled a target date of May/June 2012 for full implementation at Browns Ferry unit 1. Assuming success at the lead unit, the other two units would follow in late 2012 and spring 2013.

Areva says that XEDOR will be made available for BWR plants without further trials. However, when a decision is made to apply XEDOR to PWR plants a similar trial would be recommended in one unit before general application.

This article was originally published in the September 2011 issue of Nuclear Engineering International