Sensors and Gas Path Measurements

[finiteparts]

I just thought that I would start a thread where we could drop ideas and discuss information related to the measurements that we do in our engines.

First, here is an excellent document that gives some really good info on measurement of pressures and temperatures in gas turbines:

apps.dtic.mil/dtic/tr/fulltext/u2/a226378.pdf

From this document, I would like to suggest that others adopt the SAE station numbering as shown:





This makes it easier talking about temps and pressure when everyone is using the same number for say "Plane 4", meaning the combustor outlet temperature.  Refinement in these numbers allows more specific information such as Plane 3.0 (when they break up the planes, usually they drop the decimal and just call it plane 30) being the outlet of the compressor and Plane 3.1 being the cold side of the combustor.

Enjoy!

Chris

[finiteparts]

If you want very accurate rotor speed measurements, then the Garrett turbo speed gauge cant be beat.



Now, it isn't cheap and since the speed sensor is actually made by Jaquet, you could probably just buy the sensor and read it into a DAQ system.  It is the same one that you would use on the Borg Warner turbo speed sensor also.  See more about the pin out, etc, at Full Race...

www.full-race.com/docs/Speed-Sensor-Installation-Instructions.pdf

Micro-Epsilon also makes a really nice unit, that I have seen in laboratory use and it was very slick.  Unfortunately, it was a bit pricey, but when precision data is what you are after, it really performed.

Now, I suspect with the need to measure speeds being more and more common in sequential or VNT arrangements, I am hoping that these types of units will becoe more available and cheaper.

- Chris

[finiteparts]

"Now for something completely different."

Vibrations is definitely an area we need to focus on when building custom turbomachinery.  Small turbomachinery is particularly vulnerable to vibration issues because the rotors are often much lighter than the imbalance forces created by their very high rotational speeds.  

So, this is just a quick description of how I am going to approach monitoring the vibrations and shaft dynamics.  I am still learning, but I did have the good fortune to work with some of the industry experts and get some nice pointers along the way. 

In this first photo, I am showing the components that I am going to use to monitor the vibrational response.  The primary components are my spectrum analyzer, the charge amps for the piezoelectric accelerometers and of courseone of the larger accelerometers.  The good news?  All of these components can be found on fleaBay for relatively cheap and for the most part, they are very robust and if they are still working, they are probably still close to their last calibrated value, even if it was 30 years ago. Typical names to look for are Endevco, PCB Piezotronics, Unholtze Dicke, etc...  ( great resource!   endevco.com/Our-Resources/Literature  )

The bad news?  There is a pretty large learning curve to selecting and setting up these systems reliably.  There are tons of used accelerometers for sale, but knowing which one will work well for you frequency range and expected vibes situation is a bit of work.   Making sure that you have the proper cable, with the proper length and then the proper charge amp is even more work...but the payoff is to have lab quality measurements for really low cost.
 


So vibes are one thing, but sometimes, you need to understand what is causing the vibrational mode to be excited and usually, that is where proximity probes come in. These are extremely accurate probes that you place near the shaft to measure the rotor or shaft excursions from their nominal locations.  When these measurements are coupled to your rotordynamics predictions, you can gain more confidence that your rotor is hitting its first flex mode or experiencing a whirl instability.

So for this, I am using Kaman probes.  These eddie-current probes are very accurate and can usually measure in the 0.001 to 0.100 inch range...so they have to accurately placed.   Incorrectly placed probes, coupled with incorrect rotordynamics predictions can lead to you loosing you expensive prox probe at the best...at the worst, you damage the rotating hardware too.  As with the vibrations hardware, there is a bit of a learning curve for these too.  Proper sizing and placement is critical to getting any information at all.  Also, if you have a bad rotordynamics prediction and place the probe near a node, you will get no real response at all, even though the rotor might be going crazy an inch or two away from your probe.  So basically, I am saying that these tools exist and are extremely helpful, but they are not for the novice. 



In addition to Kaman, Philtec makes proximity probes that use fiber optics and light to measure the proximity of the shaft to the sensor.  Now obviously, these need to have a clear line of sight, so you can't put these in an oil cavity.  But they could easily be used as proximity sensors for the outer case to watch for vibrations or to see the thermal growth of a region that is getting heated.



Beyond these, there are also Bently Nevada probes on fleaBay quite often and I have had the privilege to work with those type proximity probes at work before.  They are excellent probes and there systems are really robust.  I just haven't needed to get more sensors yet and I may get some in the future.  

Now, I know that these aren't the cheapest options, but the reason I brought these up is to make some users aware of lab quality hardware that is out there, perhaps unknown to them.

- Chris

[turboron]

Chris, thanks from all of us for starting this thread. Note that proximity probes can only be used with magnetic materials. The gaps used for steel must be decreased if magnetic stainless steels are used. A proximity probe can also be used for a tachometer but you have to provide a once per revolution trigger such as a keyway. Four gaps per rev. may also be used on lower speed rotors.

Thanks, Ron

[jetric]

Hi Ron,
You can use inductive proximity sensors on non-ferrous metals this is what the sensor is on the turbo speed gauges, the sensor is mounted in the compressor housing with its tip 0.5mm away from the compressor vanes, as each vane passes by the sensor tip it breaks the magnetic flux that a coil in the tip of the sensor produces triggering the sensor.

Richard S.

Dec 23, 2020 21:16:06 GMT -5 turboron said:

Chris, thanks from all of us for starting this thread. Note that proximity probes can only be used with magnetic materials. The gaps used for steel must be decreased if magnetic stainless steels are used. A proximity probe can also be used for a tachometer but you have to provide a once per revolution trigger such as a keyway. Four gaps per rev. may also be used on lower speed rotors.

Thanks, Ron

[finiteparts]

Dec 23, 2020 21:16:06 GMT -5 turboron said:

Chris, thanks from all of us for starting this thread. Note that proximity probes can only be used with magnetic materials. The gaps used for steel must be decreased if magnetic stainless steels are used. A proximity probe can also be used for a tachometer but you have to provide a once per revolution trigger such as a keyway. Four gaps per rev. may also be used on lower speed rotors.

Thanks, Ron

Ron,

Eddie current probes can be used with any conductive material...they work well with ferromagnetic and nonmagnetic materials.  Kaman has a table of the material properties in their online documentation...here is a screenshot of the first page...

As you can see, the material capabilities are very wide.  This is the same technology that is used in the Jaquet probes use to read blade passings in turbocharger speed sensors that are capable of reading aluminum and titanium compressor wheels.

As for the Philtec probes, they are optical...obviously they do not have any requirements for the material to even be metallic.

Agreed...the air gaps are very dependent on materials for the eddie current  probes and Kaman provides a detailed description on how to calibrate the sensors to the material.  Care is required to get accurate measurements.

Now, for reading shaft excursions, we are in fact reading steel shafts, so eddie-current probes might not be mandated, but the accuracy of these probe was the driving concern when I selected them. 

If you are thinking about another type of senser, please put up information...especially if they are cheaper.

Thanks,

Chris

[finiteparts]

Dec 24, 2020 2:38:05 GMT -5 jetric said:

Hi Ron,
You can use inductive proximity sensors on non-ferrous metals this is what the sensor is on the turbo speed gauges, the sensor is mounted in the compressor housing with its tip 0.5mm away from the compressor vanes, as each vane passes by the sensor tip it breaks the magnetic flux that a coil in the tip of the sensor produces triggering the sensor.

Richard S.

Dec 23, 2020 21:16:06 GMT -5 turboron said:

Chris, thanks from all of us for starting this thread. Note that proximity probes can only be used with magnetic materials. The gaps used for steel must be decreased if magnetic stainless steels are used. A proximity probe can also be used for a tachometer but you have to provide a once per revolution trigger such as a keyway. Four gaps per rev. may also be used on lower speed rotors.

Thanks, Ron

Richard,

There a three major types of "inductive proximity sensors"...as you state, the high frequency, eddie-current sensors that are used in turbo speed sensors or manufactured by the likes of Kaman, Keyance, Bentley Nevada, etc...  But, there are also the basic magnetic-coil type and the ones that use a the change in capacitance to modify the frequency of an charge-capacitance circuit.  I only bring this up to make sure that the readers understand that if it is termed an "indcutive Proximity sensor", it doesn't necessarily mean that it is an eddie-current sensor.

Thanks!

Chris

[turboron]

All, thanks to jetric for pointing out that Garretts rpm sensor that reads on the aluminum compressor impeller blades is a type of induction proximity probe. I did a little digging and found this article which explains the differences.

Thanks, Ron

What is the difference between a selective ferrous/non-ferrous sensor and (factor 1) ferrous/non-ferrous sensors?
Ferrous (Fe) and non-ferrous (Nfe) terminology relates to inductive proximity sensors.
Inductive proximity sensors work by inducing a current in a metal object that is within the operating range of the sensor. The inductive sensor generates a magnetic field by use of an oscillator. When a metal object is close enough to be within the magnetic field, the sensor electronics detect a load change within the oscillator and the sensors signals this via the output circuit.

Several factors affect the sensing range of an inductive proximity sensor. Because of this, standard inductive proximity sensors are given a 'nominal sensing distance' (Sn) that is based on the theoretical maximum distance the sensor can detect a metal object. In practice, because of the size of the object, variation in temperature, applied voltage and type of metal, the actual sensing distance achieved is invariably less than the Sn.

A key factor in the sensing range achieved is the type of metal to be detected. Note: Inductive sensors can only detect metal.
Of particular importance is whether the metal is ferrous or non-ferrous.

A metal that contains an appreciable amount of iron is considered to be ferrous, these include; cast iron, steel and steel alloys. As the amount of iron in the alloy reduces, so the sensing distance that a standard inductive sensor can achieve reduces.
Non-ferrous metals, such as brass and copper do not contain an appreciable amount of iron and as a result detection performance of a standard inductive sensor is considerably reduced compared to Sn.

Refer to the graph below that indicates the approximate effect of the type of metal on sensing performance of a standard inductive sensor, e.g. steel can be seen to be a coefficient (km) factor of 1 (so no effect on sensing performance), whereas copper is a factor of 0.25 so the sensing distance will only be a quarter of what it would be with steel;





Factor 1 ferrous/non-ferrous sensors

The effect on sensing distance can be problematical, for instance if both steel and aluminum cans need to be detected at the same distance from a sensor. For this reason (Factor 1) 'ferrous/non-ferrous sensors' are available from Schneider Electric.
The sensing distance for these sensors is the same for both ferrous and non-ferrous materials. The Km coefficient is always 1.

Principle of operation
These sensors use a very high frequency oscillator (in the region of MHz) that performs the same with both ferrous and non-ferrous metals.

Schneider Electric part numbers;

XS1M18KPM40 diameter 18mm (version with connector also available)
XS1M30KPM40 diameter 30mm (version with connector also available)
XS7C40KPM40 Form C
Selective ferrous and selective non-ferrous sensors
Principle of operation
These sensors use the effect on attenuation of oscillation amplitude according to the metal to differentiate ferrous and non-ferrous, attenuation is significant for ferrous materials and insignificant for non-ferrous metals. It is therefore possible to design the sensor to detect only ferrous or only non-ferrous metals;

Schneider Electric part numbers;

XS1M18PAS40 (version with connector also available) ferrous version (insensitive to non-ferrous materials)
XS1M18PAS20 (version with connector also available) non-ferrous version (insensitive to ferrous materials)
Published on:23/12/2010
Last Modified on:20/11/2020
Was this helpful?

[finiteparts]

Ron,

Thanks for adding the article. Note that those sensors, unfortunately are not usable for what we are trying to do since their output is essentially binary, open/closed. These are more for detecting if a door is open or closed, a component is passing by on a conveyor belt, etc...basically, that there is some metal item within the detection range of the sensor...not really to give us feedback on how far the metal item is from the sensor.

We need the prox probes to provide a displacement value that can be recorded and later analyzed to understand the local x- and y-displacement at the sensor plane. From that we can see if the displacements trend with the rotordynamics predictions.

Additionally, these sensors are too large in diameter. Generally, manufactures state that the shaft diameter should be at least 8 X sensor diameter.

Thanks,

Chris

[turboron]

Chris, I agree. However, for my application I don't need to detect shaft movement. I am using a cheap inductive proximity probe to get a pulse per revolution for control of the load speed. I may need to go with 4 pulses per revolution due to the lower speed of 3600 rpm.

Thanks, Ron

[turboron]

All, hooking up the low cost inductive proximity sensor to read load rpm turned out to be fairly simple. My test setup include a once per revolution hole in a disc mounted on a 3000 rpm 12 volt dc motor. The key was connecting the sensor output to one of the Arduino interrupt pins (pin 2 in my setup). I notice that the rpm bounced somewhat so I averaged ten readings to smooth the output.

Thanks, Ron

[turboron]

All, I have been trying to set up a Arduino/MAX6675 system to measure EGT. I purchased a MAX6675 Breakout board and downloaded the MAX6675 header file from the internet. Code examples are available on YouTube and other sites. I was able to get good temperature measurement after adding the following code to my program.

pinMode(egt_soPin, INPUT);
pinMode(egt_csPin, OUTPUT);
pinMode(egt_sckPin, OUTPUT);

I have since learned that one of the limitations of this device is that it needs a delay of 200 milliseconds between each reading. Note that the MAX6675 breakout board has a A/D converter. The MAX31855 Breakout board is the newer improved version of the MAX6675. I have ordered this part and may use it after more research.

Also, I ordered and received an AD8495 Breakout board from Adafruit. This device has an analog output. I may use it after more research.

Thanks, Ron

[finiteparts]

Looking around, I notice that most people are not measuring their fuel flow and since this is an important parameter to have when trying to back-calculate the performance, I thought it might be good to start talking through several methods that I have used to give readers some options if they want to measure their fuel flows.

The first and easiest approach, which can be used with gaseous or liquid fuels, is a simple weight scale under that fuel tank. By recording the weight reduction of the tank and knowing the density of the fuel, you can easily calculate the fuel consumption rate.

The next ones will focus on liquid fuels.

The cheapest of these is to build a small venturi that fits in the line with upstream, throat and downstream static pressure measurements. Since it is a liquid, you can use a modified Bernoulli's equation to calculate the velocity through the venturi throat, and from the area of the throat, you can calculate the mass flow rate. I made the venturi from aluminum because it was easy to machine. The exit diffuser was reamed out with a straight flute tapered reamer with around 3 degs of taper. I used a long straight section in front of the venturi to help condition the flow. I tried it with a connected manometer, but I think regular pressure gauges would work better, even though they may need to be bleed of air before use.

The next one that I have purchased, but haven't tried yet is a variable area flow meter (sometimes called a rotameter). I bought a used and cheap Omega FL1200 series flowmeter that has a stainless "float" and measures between 4.8 and 48 gph. The hard thing is to find the right flow range and they are also calibrated for water, so when you use kerosene or whatever, you need to do calculations based on density or specific gravity to "correct" the measured flowrate to what it actually is. Now, the downside is that it has a 200 psi limit, but it may be able to go on the suction side of the pump if the pressure drop isn't too high. The spec sheet is here:
it.omega.com/green/pdf/FL1200.pdf

The third and easiest method to integrate into the data acquisition system is a turbine style flow meter, but they are usually very expensive. But, Omega is listing these:

www.omega.com/en-us/flow-instruments/flow-meters/turbine-flow-meters/p/BV1000-2000-3000-Series

with the smallest one being a good fit for my flow rates. I haven't bought one of these yet, but when I do, I will share my results.

All of these methods can be calibrated easily by using a "catch and weigh" method. Run the pump at a set pressure and stick a cup in the discharge flow for 30 second, weight the collected sample, subtract the weight of the cup and multiply by 2 and you have mass flow per minute. You can generate a mass flow verses pressure drop curve from multiple test and get a good idea of the pumps capability and your method of measure pressures or pulses per second or whatever.

I hope this helps!

Chris

[finiteparts]

Well in the search for other flowmeters, I found these two items dirt cheap on fleaBay.  The Flow Technologies turbine flowmeter can measure from 1.75 gph to 150 gph in the extended range since I got it with the RF style pickup and the SS ball bearings.  This should do nicely for the sort of stuff that I am wanting to test.



The Omega ratemeter/totalizer was completely new at a fraction of the cost of a new one.

I am looking forward to getting the opportunity to try these, but unfortunately, that may be a while since I am in the process of relocating for work.   I just wanted to get others aware so if they wanted to find similar components, at least they know what to look for.

- Chris