Natural convection based on Oberbeck-Boussinesq (OB) approximation was widely used for studying the convection patterns in different geometries. This consists of the Rayleigh-Benard Convection (RBC) and the Vertical Convection (VC). In RBC, the working medium is confined between a cold plate at the top and a hot plate at the bottom. Due to temperature difference, a buoyancy driven flow starts developing in the fluid domain. The OB assumes a constant fluid properties except accounts the linear density variation with temperature in buoyancy term of momentum equation. The two-dimensional (2D) Direct Numerical Simulation (DNS) of OB flow in a square cavity with Prandtl number (Pr) in mid-range undergoes flow reversals. The typical flow pattern is a diagonally dominant Large Scale Circulation (LSC) with two small corner rolls in the opposite diagonals. These small rolls gains energy from main diagonal circulation, grow in size and merge to form a new LSC which rotates diagonally in a direction opposite to the earlier LSC. The phase diagram of flow reversal based on Ra and Pr can be plotted based on OB study. The Non-Oberbeck-Boussinesq (NOB) effect involves inclusion of fluid property variations with temperature and additionally accounting spatial density variation in the buoyancy term of momentum equation. The majority of NOB studies both numerical and experimental were carried out with water as the working medium and observed flow reversal in certain range of Ra number. The NOB study carried at Pr = 0.71 showed flow reversals outside the OB parameter space. In the present study Rayleigh-Benard convection of Jet-A fluid by 2D DNS is performed with NOB effect. The Jet-A fluid, widely used as fuels in aero engines, has higher viscosity and lower thermal conductivity compared to water and the Prandtl number is approximately 4.5 times higher than water at mean temperature of 40°C. The fluid properties of the working fluid are defined as a polynomial function of temperature. The Rayleigh number in the range of 106-108 is considered for the study. The NOBness effect is captured in the range of 20-60K temperature difference between top and bottom plates. The Boundary layer thickness, the Centre line temperature, the time averaged temperature profiles are investigated. The Finite Volume based 2D DNS code is developed based on hybrid upwind/central differencing (HUCD) scheme for convective terms both in momentum and energy equations. For diffusion terms, the second order Central Differencing Scheme (CDS) is used for both in momentum and energy equations. The time discretization is performed by the second-order implicit Crank-Nicholson scheme. The SIMPLER algorithm is used for the pressure-velocity coupling. The MATLAB programming is used to develop the Numerical code. In order to capture the wall thermal boundary layer, a relatively finer grid is preferred with first cell height of 0.001 in the top and bottom wall normal direction, and a growth ratio of 1.01 is used for the first five layers from both the walls. The top and bottom wall distribution is adopted in the side wall direction as well. The maximum cell aspect ratio is restricted to 4 in the overall computational domain. The code is run with different grid distribution and finally 256x256 grid size is chosen for all the studies. The time step is chosen such that minimum CFL number as less than 0.25. The start-up phase of the transient analysis is carried out for 500 units and the next 500 time units are taken for statistical averaging process. The code is validated by comparing the existing results with NOB water data available in literature. Flow reversal in RBC with NOB effect is more detailed than the OB approximation. In the NOB study asymmetric corner roll growth observed against the OB case. The Jet-A fluid with higher Prandtl number undergoes flow reversals even with Ra = 108. These are the key observations in the Jet-A fuel NOB study (i) The Nusselt number (Nu) is comparable to water for similar Ra number as expected (ii) at Ra = 107, the intermediate quadrupolar flow observed between LSC reversals (iii) Ra = 108, the random flow reversals observed. In addition, horizontally and vertically nested flows with co-rotating patterns also observed. All the reversals are confirmed through vorticity variation at the domain centre and vector plots.