Passive cooling via natural circulation of gas after a loss of coolant (LOCA) accident is one of the major goals of the Gas-cooled Fast Reactor (GFR). Due to its high surface heat flux and low coolant velocities under natural circulation in post-LOCA scenarios, the capability of turbulent gas flow to remove heat from the GFR core can be impaired by either a buoyancy effect or an acceleration effect. These phenomena lead to a Deteriorated Turbulent Heat Transfer (DTHT) regime. To predict accurately the cladding temperature at the hot spot, reliable heat transfer correlations that account correctly for these effects are needed. This work addresses this need by experimentally obtaining heat transfer data and developing new heat transfer correlations that can be used in system analysis codes, such as RELAP5-3D, to reduce uncertainties of predictions in these DTHT regimes. An experimental facility was designed and built using similitude analysis to match key experimental loop parameters to the GFRs' Decay Heat Removal (DHR) system operating conditions to the largest extent possible. Through a thorough literature survey two nondimensional numbers namely (1) the buoyancy parameter (Bo*) and (2) the acceleration parameter (K[subscript v]) were identified as important indicators of the DTHT regime. The experimental data was collected for a range of (1) inlet Reynolds number from 1800 to 42,700, (2) inlet Bo* up to 1x10[superscript -5] (3) and inlet Kv up to 5x10[superscript -6]. The data showed significantly higher reduction of the Nusselt number (up to by 70%) than previously reported (up to 50%). Also, the threshold at which DTHT regime occurs was found to be at smaller non-dimensional numbers than previously reported. A new phenomenon "re-turbulization", where the laminarized heat transfer recovers back to turbulent flow along the channel, was observed in the experiment. A new single phase gas flow heat transfer map is proposed based on the non-dimensional heat flux and the Reynolds number in our data, and is shown to compare well with data in the literature. Three sets of new correlations were developed, which reflect both the buoyancy and acceleration effects and have better accuracy as well as ease of numerical implementation than the existing correlations. The correlations are based on the Gnielinski correlation and replace the Reynolds number subtracting constant by a functional form that accounts for the buoyancy and acceleration effects separately, or in the combined form through a newly introduced nondimensional "DTHT" number. The three correlation types have different complexity level, with the first being the most complex and the third being the most simple and easy to apply without any need for iterations. Additional runs with natural circulation showed that the friction factor in the DTHT regime could be significantly higher than predicted by conventional friction factor correlations, although more experiments will be needed to develop reliable correlations for pressure drop in these regimes. Overall, it is concluded that due to the low heat transfer coefficient and increased friction factor in the DTHT regime, the GFR DHR system should be ideally designed to operate outside the DTHT regime to (1) avoid reduction of heat transfer capability, (2) avoid increase of pressure drop, and (3) reduce uncertainties in predictions of the cladding temperature.