[en] As the Tritium self sufficiency is one of the major challenges for fusion reactor, breeding blankets represent one of the major technological breakthroughs required from passing from ITER to the next step reactor, usually called DEMO. One of the two blanket concepts developed in the EU is the Helium Cooled Lithium Lead (HCLL) blanket which uses the eutectic Pb-15.7Li metal liquid as both breeder and neutron multiplier. The structures, made of EUROFER, a low activation ferritic martensitic steel, are cooled by pressurized helium at 8 MPa and inlet/outlet temperature 300/500 ^oC. In this concept, the LiPb is fed from the top of the blanket and distributed in parallel vertical channels among pairs of cells (one cell for the radial movement towards the plasma, the other for the return). The liquid metal fills the in-box volume and is slowly re-circulated (few mm per seconds) to remove the produced tritium. In this paper, a local finite element modelling of the tritium permeation rate through the HCLL breeder unit cooling plates is presented. The tritium concentration in the helium circuit and remaining in the lithium lead circuit are evaluated by solving partial differential equations governing the tritium concentration balance, the thermal field and the lithium lead velocity field for a simplified 2D geometrical representation of the breeder unit. This allows estimating the sensitivity effect of coupling these different equations in order to deduce a relevant but simplified modelling for tritium permeation. This is to compare with tritium inventories studies, were the tritium permeation rate is estimated using simplified analytical modelling which generally leads to over estimate the tritium permeation rate to the coolant and so has strong influence on the coolant purification plant design. The finite element modelling performed shows that the Tritium permeation is considerable lower than the one obtained in previous estimations where nominal values of the governing parameters were used. Using, as boundary conditions, efficiency of the external purification system, this analysis can be related and integrated in the global inventory analysis. One of the main consequences for the HCLL blanket design is that the requirements of T-permeation barrier performances can be significantly relaxed to a level already demonstrated by current barriers. (author)