GEK Wiki / Ejector-(v-1)-Performance-Test
  • If you are citizen of an European Union member nation, you may not use this service unless you are at least 16 years old.

  • Finally, you can manage your Google Docs, uploads, and email attachments (plus Dropbox and Slack files) in one convenient place. Claim a free account, and in less than 2 minutes, Dokkio (from the makers of PBworks) can automatically organize your content for you.



Page history last edited by Daniel Chisholm 11 years, 11 months ago

Ejector (v 1) Performance Test

Date: conducted 2/6/09


Purpose: To determine the performance of the initial ejector implementation.


The testing aparatus: right: high P inlet manifold - air flow rotameter, propane flow rotameter, and pressure gauge. Center: ejector, swirl burner, and low P inlet air valve (1.5"). Left: Manometer on 2" inlet pipe (inlet vacuum)



When open to the atmosphere, the ejector produced a total outflow volume of ~4.2 times the inflowing air/propane mix (originally computed as ~5 when not accounting for propane in flow). This was determined finding the dilution of the outgoing gas (dividing the slope of the open best fit by the closed best fit). Closed best fit is the stociometric mix (4.68%) (no air coming in from the low pressure inlet), open best fits includes the incoming air from the low pressure inlet, showing the amount of dilution occuring.


This version 1 "design" was thrown together with available parts and not optimized. Modeling is being done to find better optimized dimensions (which will have dependancies on pump/compressor flow rate and pressure for optimization).




Ejector Test.xls



Analysis, Modelling, Validation:

(sorry, work in progress... I'll remove this "disclaimer when I finish here!  - DanielC)


In this section an attempt is made to understand the parameters that this ejector is operating under, apply these to the theoretical and engineering models that are available and produce some predicted performance, and then compare these theoretical predictions to the actual measured conditions (validation of the model).  If we can get the model agreeing with the measurements, we can then use the model to explore improvements, and provide some guidance for the next round of metal cutting.


Sometime a spreadsheet doesn't provide quite enough power and flexibility for what you would like to model; I've recently discovered "octave", and have been using it for my models and calculations that don't easily and naturally work in a spreadsheet.  It is a free (GNU) program, and has been quite useful to me so far.



An ejector consists of several parts:

  1. the motive fluid jet
  2. the pumped,or "suction" inlet fluid port
  3. a mixing section (usually straight-walled)
  4. an optional diffuser (not used on this ejector)



In this ejector, the motive fluid jet issues from a .172" orifice, which is the i.d. of a piece of standard 1/4" tubing.  Bear has told me that there isn't any special treatment on either end, it's just cut off squarely and cleanly, so I'll SWAG nozzle coefficients of 0.90 to account for this (with a rounded inlet, and an outside-chamfered outlet, we could perhaps use 0.95).  What these nozzle coefficients account for is the contraction in flow that a nonindeal entry or exit might induce.


The first thing we'll figure out is the flow rate through the jet; we are interested in both the mass flow rate, as well as the jet velocity.  Bear indicated that 9.5 psi of pressure across the tube caused 100 litres per minute of airflow. This is enough pressure such that this can't be considered "incompressible flow", so we'll have to use slightly-more-complicated equations than if we could make the simplifying assumption of incompressible flow (pressure difference less than 4psi, or flow speeds less than 150 m/s, or pressure ratios less than 1.3).

Comments (2)

bk said

at 8:59 pm on Feb 12, 2009

Thanks for working on this Daniel.

Yes, spreadsheets are limited/limiting. I figured Excel was a reasonably common denominator, which I believe can also be opened by OpenOffice for Linux/non-proprietary users.

Another interesting software project is the R statistical computing project (, which is geared towards statistics and includes its own language. It is available for Mac, Linux, and Windows. Specialized packages are downloadable, with powerful graphing capabilities. Not much focused on engineering type modelling, but should work well (I'm not an engineer though).

I used R to produce the box plots of the results from a biochar trial (, I created the tri-variate plot manually in Illustrator, but since found that there is a package for creating these graphs.

This isn't to dissuade you from using Octave though.

At some far off future point, making the developed models web accessible would be great.

bk said

at 11:26 pm on Feb 16, 2009

Checking the effect of pressure on rotameter readings, the effect is non-negligible at some of the recorded pressures.
Corrections for the important data points (no temperature correction) (using the Dwyer provided equation for temp/pressure correction):
Nominal Flow Pressure [psi] Corrected Flow [lpm]
100 9 126.974206
100 9.5 128.3066056
120 10.5 157.116881
150 31 264.478694
I'll update the spreadsheet to v2 at some point. Or better yet, run a wider set of tests on a revised ejector design, taking this into consideration/correcting for it then.

You don't have permission to comment on this page.