Investigation of refrigeration system steam ejector performance through experiments and computational simulations

Investigation of refrigeration system steam ejector
performance through experiments and computational
simulations

Recently, vapour compression using ejectors has received more interest from researchers in the air conditioning and refrigeration eld. Ejectors have the advantage of being
extremely reliable with stable operation and no moving parts leading to essentially maintenance free operation. However, ejectors have very low eciencies and this can
be attributed to the low entrained mass flow rate of the secondary stream relative to the primary stream mass
ow rate. The entrainment and mixing between primary and
secondary streams is therefore a dominant feature which requires investigation. This thesis introduces and demonstrates new methods to characterise the mixing process in a vapour compression ejector operated with steam as the working fluid. A new steam ejector refrigeration apparatus was built to visualise the mixing flow inside the ejector. New designs for the steam generator and the evaporator chamber were used in this apparatus. The evaporator chamber was developed in order to weigh the amount of water induced from the evaporator during the test time. The weighing
method was demonstrated to be suciently accurate for this application. The highest percentage dierence between the direct weighing method and a thermal method for deducing the secondary stream mass flow rate was 2.6 %, corresponding to about 0.3% of the entrainment ratio.
A CFD simulation tool called Eilmer3 (developed by the Compressible Flow CFD group
of the University of Queensland) was used to simulate the
flow within the axisymmetric
geometry of the steam ejector. The Eilmer3 computational simulations adopted
an ideal gas model for the steam throughout the ejector. The distribution of static
pressure within the ejector was accurately simulated for different primary and secondary stream conditions and different condenser pressures. The difference between
experimental data and the Eilmer3 simulations of static wall pressure was typically less than 2 %. The difference between experimental data and the Eilmer3 simulation
of entrainment ratio was 2.9% at 130� primary stream temperature and 2.6% at 120� primary stream temperature for choked ejector operation. For ejector operation at unckoked conditions, differences between experimental results and the simulations were substantial with the critical condenser pressure typically underestimated in the
simulation by around 8.8% at 270kPa primary stream pressure and 15.4% at 200kPa primary stream pressure.
To enable visualisation of the ejector flow, a transparent acrylic duct was optically designed using a ray tracing method. The machined and polished optical duct performed
as expected in that it provided a field of view over the entire height of the duct. However, the planned schileren visualisation could not proceed with the optical duct
because of shadowing effects attributed to birefringence in the duct material. Nevertheless the optical duct was successfully used to gain photographic evidence of liquid
water and ice building up within the ejector. The presence of liquid and solid forms of water within the ejector suggests the ideal gas model cannot generally be applied to
the steam ejector flows even though a good degree of success was achieved in modelling the static pressure distribution and entrainment ratio under choked flow conditions.
A new technique for exploring the mixing region generated by the steam ejector nozzle was introduced. The new approach used a pilot tube to measure the pressure profile at different positions downstream of the nozzle exit within a mixing chamber with a relatively large rectangular cross section. A special pilot probe and traversing mechanism
was designed and fabricated for this purpose. The experimental results demonstrate a shock train was established downstream of the nozzle exit. The experimental measurements indicate that pressure wave effects within the mixing jet have largely dissipated by the 70mm downstream location. Eilmer3 simulations were also used to investigate
the flow at the nozzle exit. Eilmer3 simulations duplicate experimental pilot pressure data at the first station downstream of the nozzle exit (1mm) but are not consistent
with the pilot pressure measurements at the other positions (25, 50, and 70mm).
A momentum integral analysis of a control volume in the mixing chamber was attempted. Using the Eilmer3 simulation at the nozzle exit, the momentum transport into the control volume was calculated. The momentum transport out of the control volume was the estimated using experimental data at the 70mm station based on an ideal gas analysis and an assumed constant static pressure across the jet. To reflect
possible condensation effects, values for the ratio of specific heat lower than = 1.326 were trialled in the analysis in an attempt to achieve the required momentum transport out of the control volume. However, even with
= 1.001 the momentum transport out the control volume was too high, indicating that an in flow momentum transport
contribution due to recirculation across the downstream station may be significant although it was not included in the analysis.
The schlieren method was used to visualize the steam
flow at the exit from the nozzle inside the rectangular mixing chamber which was also used in the pilot pressure surveys. The schlieren arrangement followed Toepler's method with one lens. An edge feature of the steam jet downstream of the nozzle was detected using an image analysis process applied to the video record from the schlieren visualisation. The stagnation conditions of the steam were 380kPa and 144� in this case, and the background
pressure in the mixing chamber was 3kPa.