Quantifying risk for nuclear components constructed with known manufacturing tolerances

Tritium production is of critical importance to prospective DT fusion power plants.


Tritium is not found in sufficient quantities on Earth and therefore fusion power plants will have to be self sufficient.


for every D+T Fusion reaction the breeder blankets must produce at least 1.1 tritium atoms to account for losses in the system.

The requirement for tritium

Importance of simulating tritium production accurately

Under estimating the tritium production could result excessive storage requirements. Due to the diffusivity of tritium this could also result in challenges for monitoring the tritium inventory and keeping within the tritium release limits.

Mature blanket designs such as the HCLL and HCPB achieve Tritium Breeding Ratio (TBR) values around 1.2 to 1.3 .

Tritium consumption for a 3GW (thermal) power plant is approximately 200g. Over estimating the tritium production by 1% could lead to fuel shortage of 2g per day ($60,000 per day).

Uncertainties currently considered

Research has previously been carried out on the topic of uncertainties in tritium production.


  • Uncertainty due to nuclear data [1]
  • Uncertainties due to material changes during irradiation [2]
  • Uncertainties due to over simplification of geometry [3]



[1] F. Thomas et al, Quantifying TBR uncertainty due to lead nuclear data in HCLL blanket modelling by the Total Monte-Carlo, Fusion Engineering and Design, 2017

[2] J.Shimwell et al, Spatially and temporally varying tritium generation in solid-type breeder blankets, Fusion Engineering and Design, 2015, http://dx.doi.org/10.1016/j.fusengdes.2015.04.018

[3]L. El-Guebaly et al, Toward the. Ultimate Goal of Tritium Self-Sufficiency, Fusion Engineering and Design, 2009, http://dx.doi.org/10.1016/j.fusengdes.2008.12.098

Aims of the pilot project

Quantifying the uncertainty arising from variation in the physical dimensions of breeder blanket components will allow designers to specify acceptable tolerances to future manufacturers.

This proposal seeks to establish a relationship between manufacturing tolerances and uncertainty in tritium production.

Research plan

Generate models with geometry variation

Simulate tritium production response

Generate manufacturing tolerance distributions

tolerance (mm)


tolerance (mm)


Find risk associated with tolerances and make a recommendation for acceptable tolerance distributions.

Mean =10mm, v1=11mm, v2=8mm, v3=12mm ... tritium production = 1.001

Mean =10mm, v1=13mm, v2=9mm, v3=11mm ... tritium production = 1.004

Research plan

Generate 3D CAD models of tritium breeder blankets with geometry that represents a range of different manufacturing tolerances.

*diagram shows simplified geometry, models generated will contain more dividing plates

Research plan

Simulate tritium production


This involves converting the CAD to geometry suitable for neutronics modelling.


Options include:

Constructive Solid Geometry and MCNP

Constructive Solid Geometry and Serpent

Unstructured mesh geometry and MCNP

Surface mesh geometry (STL) and Serpent

Research plan

Quantifying the uncertainty arising from variation in the physical dimensions of breeder blanket components will allow designers to specify acceptable tolerances to future manufacturers.

The simulation input parameters along with the simulation results are used as the inputs for uncertainty quantification.



























*diagram shows simplified geometry without the first wall and rear components

There is an optimal ratio of lithium ceramic to neutron multiplier.

Variation of the cooling plates' dimensions will change the volume available for pebble beds and make it difficult to ensure the optimal ratio is maintained. 

The current HCPB design for a lithium bed is 11.6mm in height. Variation in the bed height will have an effect on TBR and the volumetric heating.

Length along contour line

Constant TBR

High Risk


Lower Risk


A more informed choice can be made when selecting designs.


Design points with equal performance lie on the same performance contour line but differ in terms of associated uncertainty / risk.


Quantifying risk

By Jonathan Shimwell