Statement of the Problem

 

A key factor in the safe operation of nuclear power plant is the integrity of the reactor pressure vessel (RPV). The damage caused by neutron irradiation over the lifetime of the plant leads to embrittlement of the vessel, including the welds, with consequent effects on the operating life of the plant.

Due to the economic desirability of prolonging the operating life of reactors, the accurate prediction of the neutron fluence at the RPV and the resulting damage is becoming critical from the point of view of nuclear safety.

 

Accurate calculations could, therefore, lead to considerable financial savings if the operating life of the plant can be extended.

 

Calculational Method

 

The method which has normally been employed for this type of work has been based on the discrete ordinates code DORT, with the synthesis of three dimensional distributions from the results of one and two dimensional calculations. The main drawbacks of this method are that the geometry is restricted to XYZ and R theta Z orthogonal arrays of mesh points, and that the nuclear data is averaged over broad energy ranges.

 

To calculate neutron damage more accurately involves detailed modelling of the system and a detailed representation of nuclear data. The Monte Carlo method of calculation has the ability to reproduce the geometry accurately and has a close representation of nuclear data.

 

The application of the Monte Carlo method to calculations of neutron transport has long been recognised, but its use has become much more widespread with the rapid development of cheap computing power in recent times. Calculations of RPV fluence can be readily executed overnight.

It is widely recognised that Monte Carlo is the best method of calculation for RPV fluence but the inertia is still behind the long established discrete ordinates methods. With modern, easy to use, Monte Carlo methods there is no longer a need to compromise with more approximate methods. Serco Assurance have the expertise to help release this advantage to plant operators.

 

Serco Assurance Consultancy Services

 

Serco Assurance, a leading International Science and Engineering Company, have considerable experience with Monte Carlo analysis. Our business is solving customer problems and our traditional market is nuclear services and utilities.

 

Our track record using Monte Carlo methods includes RPV fluence analyses for PWR, BWR and Magnox reactors, giving you confidence that your assessment is in the hands of expert consultants.

We recently completed a fluence analysis for the Palisades PWR which significantly reduced the conservatisms in embrittlement data found in earlier studies. This could mean that the plant is able to continue to operate beyond its original expectations, saving $20 million - the cost of annealing the plant's reactor pressure vessel.

 

We can expertly and efficiently carry out a full Monte Carlo assessment for you, which could prove to be just as cost effective for your plant. Alternatively we could carry out an analysis over a few reactor power cycles, acting as a cost effective independent check of your main supplier.

 

The MCBEND Monte Carlo Code

 

We use the Monte Carlo code MCBEND in our analyses, Refs [1] and [2]. MCBEND is a general purpose Monte Carlo code for the analysis of neutron, gamma-ray and charged particle transport. MCBEND has been developed by Serco Assurance and BNFL and has been applied by many satisfied customers over many years. It is now used by the NRC.

 

In MCBEND the problem geometry can be modelled to whatever level of detail is required. Surveillance capsules can be modelled explicitly. The nuclear data is represented in great detail with an exact treatment of the energy loss laws. The MCBEND calculation is very close to reality and the uncertainties due to the method are minimal.

 

In RPV fluence assessments MCBEND is used to model the transport of neutrons from the core of the reactor to the RPV and beyond, giving the neutron fluence within the RPV and the resulting damage, Refs [3], [4], [5], [6], [7] and [8].

 

A MCBEND methodology topical, Ref [9], has been submitted to the NRC by Serco Assurance and is currently being considered as an approved method for RPV fluence assessments.

 

The Advantages of MCBEND over other Monte Carlo Codes

 

The perceived disadvantage of the Monte Carlo method is that it is time consuming both in preparation of the model and in execution. The developers of MCBEND have minimised both to make better use of analysts' time.

• Easier Input

The MCBEND input has been designed for users rather than developers and incorporates a range of features that speed up data input and improve the QA audit trail. One good example is the direct reading of core power maps from the output of leading reactor physics codes, such as SIMULATE, into the source input of MCBEND. This removes a massive amount of data processing and possible sources of error. Another example is Fractal Geometry, Ref [10], a powerful geometry builder that permits models to be assembled from self contained component parts. Each part can be defined in its own co-ordinate system, thus greatly simplifying construction of the complete model.

• Robust Automatic Acceleration Techniques

To achieve low statistical uncertainties in a reasonable CPU time requires some form of variance reduction technique to be applied. In MCBEND, as in other Monte Carlo codes, the standard techniques are splitting/Russian roulette and source weighting Ref [11]. However, unlike other codes, these techniques are carried out automatically in MCBEND with minimum user intervention. Just tell it where and what you wish to tally and it produces an optimal 3D importance map in space and energy, which is completely separate from the geometry model. MCBEND's automatic acceleration techniques are well tried and tested and very robust when applied to the RPV fluence problem. They are the standard to which other Monte Carlo codes aspire.

• Uncertainty Assessments

In an analysis of RPV fluence, an assessment must be made of all the associated uncertainties Ref [9]. The Monte Carlo results will have a stochastic uncertainty, which these days is small - around a few percent. Other factors which contribute to the overall uncertainty of the calculation are the uncertainties associated with the nuclear data cross sections, the dimensions of the reactor components and source data. We have all these covered in MCBEND to provide full uncertainty on the fluence.

 

MCBEND can determine uncertainties associated with the nuclear data and the dimensions of the reactor components using 'sensitivities' which predicts the fractional change in a result due to a fractional change in a given parameter - say RPV thickness. The sensitivity of the neutron fluence to small uncertainties in the basic nuclear data cross-sections, or small uncertainties in the position or size of key geometry bodies, can be determined within one calculation. To determine the uncertainties associated with the variation of source intensity and source distribution over time, separate MCBEND calculations can be run for different power distributions.

• Extensive Validation

To give confidence in the use of any code, validation evidence must be available which indicates how accurate calculations are for a given problem. This may be achieved by comparison with experimental data and standard analytical solutions, or by comparison against another computer program. Validation of a complex code such as MCBEND is based on comparisons with high quality benchmark experiments covering a range of situations.

 

Validation of the particular application of MCBEND to the calculation of neutron fluences in the pressure vessel of LWR's is provided by the comparison with the benchmarks in Refs [12], [13] and [14]. These cover penetration in water, the NESDIP2 experimental array of water and steel, and the H B Robinson radial shield for Cycle 9 of the plant's operation. The NESDIP2 experiment is a simulation of the water, core barrel and pressure vessel of a PWR shield with measurements being made within the steel slabs which represent the pressure vessel. This comparison thus provides validation of the method for points within the vessel wall as required in Ref [15].

• A Quality Product

Quality Assurance (QA) is a requirement for software which is used to perform calculations associated with the safety of reactors. QA principles embrace all aspects of a software package including development, maintenance and in-service use. In the UK these requirements led to the establishment of Serco Assurance's ANSWERS Software Service which has set up a comprehensive range of software management QA procedures covering the complete life cycle including specification, design, coding, testing and in-use support maintenance. These standards are employed in the development and validation of the MCBEND code. The Quality Management System provided by these procedures has been certified against the international standard ISO 9001.

 

The Advantages of using Serco Assurance Consultants

 

Serco Assurance is an independent Science and Engineering Company with over 23 locations world-wide. It employs over 3500 people, over 2200 of which are graduates or technically qualified people.

Serco Assurance is at the forefront of research and development within the nuclear industry and has gained considerable practical experience through the operation of a number of reactor plants. This enables us to offer specialist consultancy services provided by teams of experts who have direct access to our wide range of software, to solve a wide range of customer problems.

 

Our consultancy service is unbiased and independent of regulators, plant operators and equipment. It is characterised by integrity, high technical credibility and internationally recognised authority. Our expertise has been successfully applied to provide an authoritative peer review service.

 

Quality is of prime importance in our work, both our Software and the working standard of our staff are certified against the ISO-9001 BS5750:Part 1 Quality Standard.

 

Serco Assurance is a major contributor in the field of Monte Carlo development and analysis, particularly in the area of RPV fluence assessment. We participated in the LWRSDIP programme in the '80s where we provided international benchmark experiments, such as the Winfrith Iron Benchmarks, PCA Replica and the NESDIP PWR Radial Shield series of experiments. These, together with other experiments from Serco Assurance, form a significant part of the RSICC benchmarks.

 

More Information

 

For further information on our RPV Fluence Assessment Product, contact Caroline Middlemas, telephone #44 (0) 1305 851191.

 

References