logo

Same 2-inch wheel: R245fa 3.9 kW vs. CO2 70 kW

Infinity Turbine · CO2 Turbines
ENGINEERING ANALYSIS · Rev A · 15 Nov 2026

Same 2-inch wheel. 3.9 kW on R245fa, 70 kW on CO2.

Infinity Turbine's 2012 inflow radial turbine rotor was drawn for R245fa. Run the numbers with CO2 at the same 100 °C heat source and 10 °C cooling, and the same 50.8 mm wheel becomes a 50–70 kW-class machine. This analysis shows why, what it costs in pump power, and how the wheel behaves on a small CO2 lab cart.

3.9 kWR245fa · 0.13 kg/s · 75k rpm
vs
69.6 kWCO2 120 bar · 2.8 kg/s · 68k rpm

Measured geometry · meanline method · wheel capacity · partial admission · expander comparison · rotor stress · recommendations

Render of the 2012 Infinity Turbine 2-inch inflow radial turbine rotor with 10 full blades and 10 splitters
2012 inflow radial rotor: Ø2.000 in × 0.750 in, 10 blades + 10 splitters, 304 stainless.
18×more shaft power on CO2 (120 bar) than on R245fa from the same wheel
24×denser exhaust with CO2, so far more mass passes the same passages
~16×more flow than the Cat 310 pump can deliver at the wheel's natural point
2.4 kWbest net shaft on the lab cart at 4 gpm (about 32k rpm, 43 %)
Bottom line

A well-proportioned wheel, sized for much more flow than a lab cart

This is a conventional 2-inch radial inflow turbine wheel. With CO2 at its natural operating point it would pass about 2.8 kg/s at about 68,000 rpm, roughly 16 times what the Cat 310 pump can supply.

On the lab cart it can only run in deep partial admission at reduced speed: about 2.4 kW at 4 gpm and about 150 W at 20,000 BTU/hr. Running it at its CO2 design speed on the cart would cost more power in churning the idle blade passages than it produces.

  • Original R245fa duty: about 3.9 kW shaft at 0.13 kg/s and about 75,000 rpm.
  • CO2 at full flow: about 48 kW at 100 bar or 70 kW at 120 bar. It needs 0.55–0.62 MW of 100 °C heat and a 2.3–2.8 kg/s pump using 15–26 kW.
  • On the lab cart: the nozzle covers only 1–4 % of the rotor circumference. Best speeds are 15–32k rpm, at 23–43 % efficiency.
  • Versus other cart expanders: below the bucket turbine at 4 gpm, and below the Tesla stack at 20,000 BTU/hr.
  • Mechanical: at 68,000 rpm the bore stress in 304 stainless reaches yield. Use 17-4PH H1025 or Ti-6Al-4V for full-speed CO2 work. At 25–35k rpm, 304 is fine.
1 · The rotor

What the 2012 files show

Geometry was measured by slicing the 2012 STL model every 0.02 in and checked against the 2013 ITmicro Radial Impeller drawing, which has the same Ø2.000 × 0.750 in envelope in 304 stainless.

Meridional half-section of the rotor measured from the STL, showing hub, blade passage and shroud line
Figure 1. Meridional profile measured from the 2012 STL (half-section).
2013 ITmicro Radial Impeller drawing with front, side and section views
Figure 2. 2013 ITmicro Radial Impeller drawing (front, side and section views).
DimensionImperialMetricSource / note
Wheel / back-disc diameter2.000 in50.8 mmSTL and 2013 drawing
Rotor inlet (blade tip) radius r10.975 in24.8 mmSTL slices
Inlet blade height b10.12 in3.0 mmSTL, back face to shroud break
Back-disc thickness0.063 in1.6 mmDrawing
Exducer hub / tip radius0.33 / 0.59 in8.4 / 15.0 mmSTL at z = 0.70 in; r3s/r1 = 0.61
Exit annulus area (gross)0.751 in²4.85 cm²About 12 % blade blockage assumed
Inlet passage area (gross)0.735 in²4.74 cm²2π r1 b1
Overall axial length0.750 in19.05 mmSTL and drawing
Blades10 full + 10 splittersSplitters end about 0.57 in from the back face
Inlet blade angleAbout 0° (radial element)Planar wrap under about 10°
Exducer exit blade angleAbout 60–62° from axialBlade wrap 4–5° per 0.02 in near the exit
Blade thickness (exducer)About 0.03 inAbout 0.8 mmSTL sections
Bore / counterboreØ0.265 thru / Ø0.394 × 0.625 deepØ6.7 / Ø10.0 mmDrawing
Volume / mass (304 SS)0.934 in³121 gSTL volume

The STEP file path ("CO2T_LCO2 turbine with pump") and its name ("R245 Inflow Radial Turbine") suggest the wheel was drawn for R245fa and considered for CO2.

2 · Method

One-dimensional meanline, real-fluid properties

Fluid properties come from CoolProp 8.0 (Span–Wagner for CO2, Lemmon for R245fa). The same cycle basis is used for both fluids so the comparison is like for like.

  • Cycle: 100 °C source, turbine inlet 90 °C; 10 °C cooling water, so CO2 condenses at 18 °C (54.7 bar). R245fa: saturated at 85 °C (8.9 bar), 5 K superheat, condensing at 18 °C (1.14 bar).
  • Design speed: classic radial-turbine optimum, blade-tip speed U1 = 0.70 × spouting velocity c0.
  • Flow capacity: set by the exducer, exit axial velocity C3 = 0.25 U1. This gives specific speed Ns = 0.41 for every fluid, with peak efficiency about 78 % from a radial-turbine Ns correlation.
Partial admission on the cart. Lab-cart flows (Cat 310 at 1–4 gpm, and the 20,000 BTU/hr case) are far below capacity, so the wheel is modeled in partial admission. The nozzle arc follows from nozzle exit density, velocity, blade height and a 75° nozzle angle. Losses: a stage-efficiency curve peaking at 70 %, ventilation of idle passages ζv = 0.13 (1−ε)/ε (U/c0)³ (Suter–Traupel type), and back-face disc friction (Daily–Nece).
Uncertainty. About ±10 efficiency points on the meanline results, and about ±50 % on ventilation loss, which matters most at high speed. The trends and the factor-of-16 flow mismatch are robust.
3 · Wheel capacity

CO2 versus R245fa at the wheel's natural operating point

With CO2, the same 2-inch wheel produces 12–18× the shaft power it makes on R245fa. The cost is a much larger pump: back-work is 30–37 % for CO2 versus 2 % for R245fa.

At the natural operating pointR245fa (design fluid)CO2, 100 barCO2, 120 bar
Turbine inlet8.9 bar / 90 °C (129 psia / 194 °F)100 bar / 90 °C (1,450 psia)120 bar / 90 °C (1,740 psia)
Pressure ratio7.81.812.17
Isentropic enthalpy drop38.6 kJ/kg27.2 kJ/kg31.5 kJ/kg
Spouting velocity c0 / nozzle Mach278 m/s / 1.9233 m/s / 0.95251 m/s / 1.10
Design speed (U1/c0 = 0.70)75,008 rpm (195 m/s)62,940 rpm (163 m/s)67,766 rpm (176 m/s)
Exhaust density6.3 kg/m³130 kg/m³150 kg/m³
Mass flow capacity0.13 kg/s (17 lb/min)2.26 kg/s (298 lb/min)2.82 kg/s (373 lb/min)
Specific speed Ns / est. efficiency0.41 / 78 %0.41 / 78 %0.41 / 78 %
Shaft power3.9 kW (5.3 hp)48.0 kW (64 hp)69.6 kW (93 hp)
Heat input required33 kW (111,524 BTU/hr)548 kW (1.87 M BTU/hr)623 kW (2.13 M BTU/hr)
Pump work (fluid)0.09 kW14.6 kW26.1 kW
Net electric (95 % generator, 85 % pump drive)3.6 kW28.4 kW35.4 kW
ρ

Density does the work

CO2 leaves the turbine about 24× denser than R245fa, so far more mass flows through the same passages.

M

Easier nozzle

The R245fa nozzle is supersonic (Mach 1.9). The CO2 nozzle is near sonic (Mach 0.95–1.10).

P

Bigger pump

CO2 back-work is 30–37 % of shaft power versus 2 % for R245fa, so net electric is 28–35 kW, not 48–70 kW.

4 · On the CO2 lab cart

Deep partial admission at reduced speed

At cart flows the nozzle arc is less than one blade passage wide (inlet blade pitch is 7.8 mm). The idle passages churn dense CO2 at about 150 kg/m³, so the best speed drops well below the design point.

Flow caseCO2 flowAdmitted arcNozzle areaBest speedIsentropic powerNet shaftEfficiency
20,000 BTU/hr (24 g/s)0.024 kg/s (3.2 lb/min)0.7 % (1.0 mm)0.8 mm²15,000 rpm0.66 kW0.15 kW23 %
Cat 310 1 gpm0.044 kg/s (5.9 lb/min)1.0 % (1.5 mm)1.2 mm²19,000 rpm1.40 kW0.38 kW27 %
Cat 310 2 gpm0.089 kg/s (11.8 lb/min)1.9 % (3.0 mm)2.4 mm²25,000 rpm2.80 kW0.97 kW35 %
Cat 310 3 gpm0.133 kg/s (17.6 lb/min)2.9 % (4.5 mm)3.5 mm²29,000 rpm4.20 kW1.66 kW39 %
Cat 310 4 gpm0.178 kg/s (23.5 lb/min)3.8 % (6.0 mm)4.7 mm²32,000 rpm5.61 kW2.41 kW43 %

Cat 310 cases at 120 bar; the 20,000 BTU/hr case at 100 bar.

Net shaft power versus rotor speed for each lab-cart flow case, with peaks between 15k and 32k rpm and the 68k rpm CO2 design speed marked
Figure 3. Net shaft power versus speed. Every curve turns negative well before the 68k rpm design speed.
Stacked bars showing net shaft power, ventilation loss and nozzle, rotor and exit losses at each flow case
Figure 4. Where the isentropic power goes at each flow (best speed; label = efficiency).
Why speed hurts here. At low admission the best speed falls to U/c0 of 0.2–0.33, far below the 0.7 the wheel was designed for. Ventilation rises with the cube of speed: at 68,000 rpm the idle passages would absorb about 6 kW, more than any cart flow can supply.
5 · Comparison

Against the other lab-cart expanders

Left: shaft power at full flow capacity, 3.9 kW R245fa, 48.0 kW CO2 100 bar, 69.6 kW CO2 120 bar. Right: lab-cart expander comparison, ideal 5.61 kW, bucket turbine 3.65 kW, this wheel 2.41 kW at 4 gpm
Figure 5. Left: the wheel at its full flow capacity. Right: expander options on the lab cart.

At 4 gpm, this wheel (about 2.41 kW) is below the bucket turbine estimate (3.65 kW at an assumed 65 %). At 20,000 BTU/hr it is below even the Tesla stack (about 0.15 kW versus a 0.29 kW model ceiling).

On the cart, this wheel is a reasonable demonstrator at 3–4 gpm but not the best performer. Its strength is at much higher flow.

Bucket turbine and Tesla figures are from the earlier lab-cart and Tesla reports in this series.

6 · Mechanical notes

Rotor stress versus speed

SpeedTip speedNominal centre stressAt bore (×2)304 SS (yield about 205 MPa annealed)
32,000 rpm (cart, 4 gpm)83 m/s22 MPa45 MPaFine
50,000 rpm130 m/s55 MPa110 MPaAcceptable
68,000 rpm (CO2 design)176 m/s101 MPa203 MPaAt yield: use 17-4PH H1025 or Ti-6Al-4V
75,000 rpm (R245fa design)195 m/s125 MPa250 MPaOver yield

Stresses use the rotating-disc approximation (3+ν)/8·ρU², doubled at the bore.

Before any high-speed CO2 run

Size the rotor and bearings for full admission

  • Material: at the 68,000 rpm CO2 design speed, 304 stainless reaches yield at the bore. Re-make the wheel in 17-4PH H1025 or Ti-6Al-4V for any run above about 50k rpm.
  • FEA: blade root and splitter stresses need a proper finite-element check before any high-speed run.
  • Balance to G2.5 or better.
  • Thrust: with CO2 the back face sees roughly nozzle-exit pressure and the exducer sees exhaust pressure. Size the axial thrust bearing for the full-admission case even if the cart runs in partial admission.
  • Pressure: CO2 turbine inlets here run at 100–120 bar (1,450–1,740 psia). Casing, seals and piping must be rated accordingly.
7 · Recommendations

What to do with this wheel

  1. On the lab cart (Cat 310, 1–4 gpm): run one converging nozzle of about 4.7 mm² (4 gpm) to 1.2 mm² (1 gpm), set about 75° from radial, at 25–35k rpm. Shroud the non-admitted arc to cut ventilation; this could add several hundred watts. Expect about 1.5–2.5 kW at 3–4 gpm.
  2. Skip it for the 20,000 BTU/hr case (about 0.15 kW). A positive-displacement or hydraulic expander, such as the ROT 06 concept, or a much smaller impulse wheel suits that flow better.
  3. As a production machine, this wheel suits CO2 at about 2.3–2.8 kg/s: 48–70 kW shaft from 0.55–0.62 MW of 100 °C heat. That is a natural single-module size for the Cluster Mesh data-center waste-heat concept, with a pump in the 15–26 kW class.
  1. Baseline with R245fa: run the wheel on the IT10 at its design fluid (about 3.9 kW at 0.13 kg/s) to get a measured efficiency that calibrates the CO2 predictions.
  2. Above about 50k rpm on CO2: re-make the wheel in 17-4PH or titanium, check blades by FEA, balance to G2.5 or better, and design the thrust bearing for CO2 pressures.
  3. Scaling is the problem at cart flows. Holding the same specific speed at 0.18 kg/s would need a 12 mm wheel at about 270,000 rpm. That is why small CO2 test loops favor partial-admission impulse or positive-displacement expanders.
Get started

Report, CO2 turbines and project help

CO2 Duty Analysis Report

PDF
  • 7-page engineering report
  • Measured rotor geometry and drawings
  • CO2 vs R245fa capacity tables
  • Lab-cart partial admission, stress and recommendations
Download PDF

50–70 kW CO2 module

Data-center waste heat
  • 2.3–2.8 kg/s CO2 per module
  • 0.55–0.62 MW of 100 °C heat
  • Cluster Mesh concept sizing
Ask about a module

Turbine consulting

Quote
  • Radial and impulse turbine sizing
  • CO2 and ORC cycle analysis
  • Test loop and expander selection
Ask about a project
FAQ

Common questions

Why does the same wheel make so much more power on CO2?

CO2 at the turbine exit is about 24× denser than R245fa, so the same passages carry about 20× more mass flow (2.8 kg/s versus 0.13 kg/s). The enthalpy drop per kilogram is a little lower, but the mass flow wins: about 70 kW versus 3.9 kW shaft.

Is the 70 kW net electric?

No. 69.6 kW is shaft power at 120 bar. The CO2 pump takes about 26 kW of fluid work, so net electric is about 35 kW with a 95 % generator and 85 % pump drive. At 100 bar it is 48 kW shaft and about 28 kW net.

Why not run it at full speed on the lab cart?

The cart's Cat 310 pump supplies about 1/16 of the wheel's flow capacity, so the nozzle covers only 1–4 % of the circumference. The idle passages churn dense CO2, and that ventilation loss grows with the cube of speed: about 6 kW at 68,000 rpm, more than any cart flow can supply.

Can the 304 stainless wheel run at CO2 design speed?

Not safely. At 68,000 rpm the bore stress is about 203 MPa, at the yield of annealed 304. Use 17-4PH H1025 or Ti-6Al-4V, verify by FEA and balance to G2.5 or better. At the cart's 25–35k rpm, 304 is fine.

What expander suits a small CO2 test loop?

At cart flows, a radial wheel with the same specific speed would be 12 mm across and spin at about 270,000 rpm. Partial-admission impulse wheels or positive-displacement expanders are the practical choice there.

Small wheel. Dense fluid. Big output.

The same 2-inch rotor goes from 3.9 kW on R245fa to 70 kW on CO2, given the flow and heat to feed it.

Download the report

Questions? Email greg@infinityturbine.com

CONTACT Email: greg@infinityturbine.com | AMP | PDF