【正文】 ement time predicted by power law scalings [11]: IPB98(y,2) for ELMy Hmode and standard power law regression for Lmode. The EC power deposition profile was calculated by the TORAY raytracing code.. High density ELMy Hmode regimeThe target parameters for modelling were taken from Ohmic and X3 heated (Table 1, No. ) stationary ELMy Hmode phases of TCV discharge [12]. About 95% of injected 25keV deuterium NB power can be absorbed by the plasma for tangentially injected beam. The simulations show that Ti0=Te0? can be achieved with ～ NBI and (Figs. 1 and 2).Access to Ti/Te 2 should be attainable at increased (2MW)NB or reduced X2ECH power. The fast ion chargeexchange (CX) losses on background neutrals strongly depend on the first wall recycling conditions, the density of background atoms n0DLCS=51015m3 ,obtained from EIRENE modelling, reduces the NB heating efficiency by ～15% (No. ), CX losses on beam neutrals (≤2%) are neglectable.At high plasma density and current, neutral beam injection could result in an increase of the thermal βN from (pure MW X3ECH) to (2MW NBI), and could even reach the ideal MHD limit (～3) resulting from the fast particle contribution. Fast ion slowing down times in such regimes are of the order of 10 ms,. shorter or parable with the bulk plasma energy confinement time, so, perturbation of the ion energy Maxwellian distribution by fast ions is expected to be small (as in a fusion reactor).. X2EC and NBI heatingModelling of NB heating in low density regimes was performed for 2MW X2EC heated Lmode reference discharge (31761, ). Increase of the NB deposited power per plasma ion at low density results in ～2 times lower (≤) than in high density regime NBI power required to access Ti0 of 2~3 keV (scenarios and and Fig. 3). Nearnormal NB injection (ENB ≥15 keV) cannot be considered here due to higher (20%) shinethrough losses, resulting in first wall overheat of the TCV central column. ASTRA simulations confirm earlier experimental and numerical studies of fast ion orbit losses on the TCV [13]. At low plasma current, fast ion orbit losses are extremely important and bee substantial for counterIp NB injection (Fig. 4)。 losses increase at high ion energy (32% for 25 keV DNB and 59% for 50 keV, scenarios and ) and for higher NB atomic mass. NB injection at low plasma density and current provides the possibility to study the fast ion and MHD physics. In the unfavourable scenario (like ), the 200~300kWdelivered by the NB power leads to the creation of a strong fast ion population with a stored energy of few tens kJ that, at low current, significantly contributes to the ideal MHD ˇ limit. Fast particle instabilities would dominate the plasma behaviour under these conditions [5]. beams injection layoutTCV was not originally designed for neutral beam heating although several relatively wide machine midplane lateral ports were implemented for general diagnostic flexibility. The location of magnetic field coils, for which modification is not feasible, and the existing support structures are major problems for NBI plasma access, in particular for the tangential injection direction. Access for NB injectors through 15cm diameter portswith near normal injection (tangency radius Rtan ≤23 cm) and through a single 216。10cm aperture neartangential injection port with the axis passing near the inner wall at Rtan? 65cm has been analysed in [13]. Shine through for Rtan ?23 cm is workable at the high densities。 NB usage at low densities is, however, severely limited by excessive shinethrough and high inner wall power loads. The maximal acceptable power load of [14] of 1000K corresponding to～10% shinethrough of the 1MW beam with the 15cm footprint size. A model of a neutral beam with geometric focussing and angular divergence [15] was performed to calculate the beam transmission and power load on the critical scrapers in the NBI duct. The acceptable ～80% beam power transmitted into the tokamak for 1MW, 25 keV, 1 s beam with 200mA/cm2 extraction current from the ion optical system located at about 250cm from the TCV port is feasible only with low beam divergence: 。for 216。10/15cm duct apertures respectively. The transmission of the high power (~) NB through narrow ports demands high current density, low divergence neutral beam injector only reachable, at present, by lower current diagnostic neutral beams. To allay these requirements on beam divergence and current density a modification TCV vacuum vessel to create new port(s), specifically designed for NBH and fitted between magnetic field coils, is considered. The available gaps between toroidal and poloidal magnetic field coils at the TCV midplane are 22cm in vertical and 38cm in toroidal direction. The design of duct with innerminimal aperture of 20 cm, wall thickness 1 cm and 3 cm gaps to toroidal field coils, beam axis tangency radius of 74cm (Fig. 5) was found to be feasible and permits to transmit 90% of the NB power to the plasma for 1MW, 25 keV deuterium beam with divergence≤。 (reachable for heating beams). The relation between Rtan and beam duct aperture horizontal size for chosen duct wall thickness and gaps to toroidal coils is shown in Fig. 6. To reduce beam blocking by desorbed gas in the narrowest part of the beam duct (close to the tokamak entrance), differential duct pumping is required. This geometry could permit two NB injectors (aiming in co and countercurrent directions) on the same port. With proper power adjustment, one could obtain scenarios with balanced momentum transfer to the plasma.4. ConclusionInstallation of 1MW, 25 keV, deuterium, tangential (basic reference) neutral beam injector would significantly increase the experimental capability of the TCV tokamak by extending the operational domain at higher Ti/Teratio and plasma pressure (β) and widening Hmode operational domain (especially at high density). 1MW of injected power is sufficient to access Ti/Te2k