Thermofluids laboratory programme

The programme

I led our group of five. I allocated the work, planned it on a Gantt chart in ProjectLibre, and was the group’s contact with the lecturers and lab technicians. In every experiment I took readings and worked through the calculations with the group, and checked our results with the supervisors. We wrote one report, with a section each:

Fluid flow (my section)

Air-flow measurement in a duct.

Energy conservation

A power balance on a heated wind-tunnel rig.

Refrigeration

A vapour-compression cycle at two evaporator loads.

Heat exchanger

Co-current and counter-current flow in a concentric-tube exchanger.

Archimedes’ principle

Buoyancy, and the densities of irregular samples.

Fluid flow: what I did

The rig is a fan-driven 140 mm air duct with three measuring devices: a Venturi tube with an 89 mm throat, an orifice plate with a 108 mm bore, and a Pitot tube in a 127 × 115 mm section. Pressure differences were read on inclined manometers set to a ×0.2 multiplier.

1. Levelled and zeroed the manometers, and recorded the multipliers.
2. Recorded the ambient conditions, 1,033 hPa and 19.6 °C, giving an air density of 1.23 kg/m³.
3. Ran the fan with the valve opened 15 mm and read the pressure difference across the Venturi and the orifice plate, and at the Pitot tube in two positions.
4. Repeated at 30 mm and 50 mm.

Opening the valve cut the system’s resistance, so the flow and every pressure difference rose. The two Pitot positions read different speeds because wall friction slows the air near the duct walls.

(opens full size)

(opens full size)

The group’s other results

Energy conservation

At full power the calculated output (1,942 W) came out higher than the 1,840 W electrical input. The group traced this to the centreline pitot tube reading peak rather than mean velocity, which overstates the mass flow. The uncertainty in total power was 8–15%.

Refrigeration

Raising the evaporator load from 30% to 80% doubled the refrigerant flow, from 2.54 to 5.16 g/s, and raised the evaporator power from 146 W to 892 W.

Heat exchanger

All four runs were fully turbulent, with Reynolds numbers of 25,000–38,000 and a Prandtl number of about 3.3.

Archimedes’ principle

Measured buoyant forces against the weight of displaced water, and the densities of irregular samples from the water they displaced.

(opens full size)

Looking back

The measurements in my section are sound, but my analysis stopped one step short of the experiment’s aims.

• Two of my “velocities” weren’t velocities. I applied the Pitot formula, v = √(2ΔP/ρ), to the Venturi and orifice readings as well. Those devices need the area ratio and a discharge coefficient: Q = C_d·A₂·√(2ΔP / ρ(1 − β⁴)). Even with C_d = 1, the Venturi reading at 50 mm gives a mean duct speed of about 14 m/s, not the 31.7 m/s my table implied.
• I didn’t calculate the volume flow rate. That was the first aim, and my conclusion said the Venturi and orifice gave reliable flow-rate data without a single Q value to back it.
• Two points aren’t a velocity profile. A traverse needs perhaps 7–10 stations across the 127 mm section, and integrating it would have given a third, independent flow rate to check the other two against.
• The fan’s performance was never measured. The third aim was the effect of load on the fan, but without its pressure rise and torque there’s no fan curve, only a qualitative statement.
• There was no uncertainty analysis and no repeats. If the scale reads to the nearest millimetre, the smallest Pitot reading (33 Pa) carries about ±6% from that alone.
• “The Venturi is more efficient than the orifice plate” is textbook knowledge, not something we measured; we never read the permanent pressure loss downstream of each device. The standard I cited, ISO 5167, doesn’t strictly cover this rig either: the orifice’s diameter ratio (0.77) is above its 0.75 limit, and the duct’s Reynolds number is below the range for its Venturi coefficients.

As group leader I’d also tighten the rest of the report. Our refrigeration section never calculated the COP or compressor efficiencies, and the heat-exchanger section never calculated the LMTD or heat-transfer coefficient, although both were aims. In the heat-exchanger runs, the two flow arrangements gave the same heat transfer: what theory predicts at these conditions, but not the “counter-current is better” result we concluded.

What I’d do differently

• Calculate Q for all three devices with the brief’s discharge coefficients, and plot flow rate against valve opening with error bars.
• Run a full Pitot traverse and integrate it as an independent check.
• Record the fan’s pressure rise and torque to draw a fan curve.
• Repeat each reading and carry the uncertainty through every step.