[Gerry Kan]

Folks:

I would like a quick opinion on this. I would like to measure continuous wind in an urban environment (e.g., on the roof or between buildings) for at least a month at a time up to six months. My probe is a weather station that can give me horizontal wind speed and direction but at 1 Hz.

Given the typical wind speed (about 4 m/s or less), would it be reasonable to assume that, at this sampling frequency, my wind measurement at least cover the inertial range reasonably well?

By that vein, if I already cover the inertial range of the spectrum, would I be able to obtain a reasonable turbulent velocity fluctuation (and hence the turbulent intensity)? Is this kind of why TI is used for evaluating CFD models (instead of directly comparing fluctuations?)

Thanks in advance, Gerry.

P.S. - I know this who discussion also depends on my averaging window (i.e., I can always cover the "inertial range" if my time window is "large enough"). I just want some opinions of one who have done similar measurements.

[LuckyTran]

The sampling frequency may not, and usually is not, the same as the actual frequency of the signal content. You may signal the anemometer at 10 kHz by reading voltages faster but that doesn't mean the anemometer provides this type of frequency resolution.

The weather station uses some kind of anemometer, usually it is a a small rotating vertical axis turbine that measures torque which they then correlate/calibrate to general wind speed. Even if the weather station data is available at 1 second intervals, the response speed of the turbine must be sufficiently fast to respond to changes in wind speed at higher frequencies. Furthermore, it is an omnidirectional wheel that measures fluctuations in velocity magnitude so it has no clue whether the velocity fluctuation is u', v' or w' or what combination of these. Hence, turbulence intensity is the much more practical quantity and it might not actually even be the turbulence intensity that is measured. You can imagine that after decades nobody has thought to reapply oil lube to the turbine bearings and there is no frequency response anymore. The anemometer at the weather station most likely reads the mean velocity signal and doesn't contain any info about the turbulence and only the unsteadiness in the wind.

What you need to provide to two equation models is tke. TI let's your determine k very quickly and scales with the flow velocity. RSM models instead need the reynolds stresses uu,uv,uw, vv, ww. Here you need covariances. Since your anemometer has no clue what components it even measures, you don't gain any ground here. Hence, TI is the way to go when you have no clue what it is you are even looking at.

[FMDenaro]

I don’t think your device can produce useful measurements, your frequency response is too low.

Do you have some multidimensional probe?

[Gerry Kan]

Thanks Folks, and nice to talk to you again:

I kind of got the sense that's why TI is used (in some other real-world / wind tunnel measurements): the frequency response is too low to capture all possible turbulent scales. I also am in the opinion that, if you have to use TI, the TI itself will, politely put, have a huge error band around it. On this point I and glad that you two also have a similar opinion.

The weather stations I have is a little more sophisticated than a vane anemometer; it's a sonic sensor (no moving parts) that will spit out speed and direction for horizontal wind (i.e., no w component).

As an aside, what kind of bothers me is that, since it's sonic, it would have natively measured u,v. But being a weather station, they were converted to speed and direction instead.

Given that it is for long-term observations, there are other practical concerns (i.e., missing data, power consumption, storage, and vandalism) that prevents me from contemplating (let alone deploying) expensive probes that gives me, say velocity component readings at 1 kHz.

So - in other words - as cursory gesture I should at least at an attempt to compare TI between the measurements and my CFD results.

Gerry.

[LuckyTran]

Acoustic flowmeters measure the sound speed which is a line/volume-integral convolution of the local temperature which generally also does not give any information about turbulence. For sure if you want turbulence measurements, then you need to look into the instrumentation and their capabilities.

It is a simple task to convert u,v,w into wind speed and direction and vice versa. You just invert the same excel functions used. It is a mathematical identity. This point should not bother you at all. Vectors are the same whether you refer to them by their components or by magnitude and direction.

Probes are not necessarily expensive if you are trying to install one at a specific location in mind. You can buy hot-films in bulk for $100-200 a piece. They are probably 10x cheaper than the robust industrial quality anemometers used to measure daily wind speed. It's like buying furniture. You can get a folding chair for $2 to use in your backward but if you are a "business" then the same folding chair suddenly costs $200. Nonetheless I understand your task is not instrumenting a weather station. It can be done, but the takeaway is that commonly used instruments don't measure any turbulence.

[Gerry Kan]

Hi Lucky:

I have seen sonic probes that can operate up to 100 Hz, though they are huge and expensive. The more reasonable sized ones should give you 50 Hz. Okay, it's still not good enough for lab or wind tunnel conditions but it will cover the inertial subrange reasonably well. Again, a trade off between ruggedness and precision.

On the other hand, mine isn't that sophisticated; it can sample at a measly 1 Hz. But then again, the rig is intended to measure (horizontal) mean wind.

Hot wires are cheap but they won't be able to take the abuse (it's outdoors and continuous). But if there is anything constructive that comes out of this conversation: The more I think about it, the more I think I need to shell out money for a proper probe that I know will work hassle free.

Gerry.

[LuckyTran]

The bigger the probe the slower its response time because it has a large mass! How fast it outputs data means nothing! DAQs nowadays operate in a millions of Hz. Anyone can hook up a pitot probe and sample it at 10 MHz, that doesn't mean you suddenly measure Kolmogorov eddies with a pitot probe

[Gerry Kan]

No, no ... the 100 Hz I am talking about is the output frequency of the probe, not the number of samples the DAQ reads in.

Boy are these things expensive ... almost 10k a piece. Look like I will have to shell out that money.

[LuckyTran]

Let me say it again! The response of the electronic signal that comes out of any device is not a direct indicator of the response it can measure. Measurement devices have been digital for a long time and use all sorts of converters that can operate at any frequency, whatever the electric engineer decided to cook up. The frequency of the electronics is what the probe can actually measure.

In general you do not want a "flowmeter" because these are volume integrated devices. The turbulence intensity in a pipe flow for example can easily exceed 50% near walls but an ultraosonic flow meter will not see any of that at all. It sees a volumetric pulsation of nearly 0%.

You need to talk to someone that knows how measurements works, talk to the engineer that made the sensor for example. Or just post the specs of your probe and we can roughly explain to you how it works.

[Gerry Kan]

Then I think you need to look up how a sonic anemometer works.

Also, by your argument, all probes are volume integrated. The question is what the volume is, and how big it is in relation to the extent of your region of interest. I am measuring a large volume (i.e., open space) and the integrated volume will have to be large. I must admit the measuring frequency is less than what I am used to (hot wire / wind tunnel) but this comes with the territory.

As an aside, based on the way you replied to this thread I got an inkling you are took me as some kind of idiot. If you really think these kind of questions are below your intellect, you could elect not to answer, and I respect your opinion and would not get offended. Just don't tell this in front of my face.

I tried to be cordial and patient in my response but they were quite obviously mistaken as weakness.

[LuckyTran]

My point was that "flow meters" intentionally and physically volume integrates a flow by adding flow conditioners. An ultrasonic flow meter for example, or mass air flow sensor used in an automobile are both similar to sonic anemometers and "hot wires" but they straighten out the flow before measuring them. Vane anemometers typically used by weather stations are kinda the same idea. So make sure you don't have a flow meter but an open instrument. Btw if you were looking for cheap hot films, you could probably pull them by the thousands from a car junk yard. These are designed to withstand abusive drivers.

Okay so you are sure your sonic sensor works all the way up to 100 Hz? Which means it has a sub-milisecond time constant? Again, all you had to do was post the spec and there wouldn't have been any confusion. Still, a large device doesn't help. You'd rather have a small one if you want more temporal resolution. Furthermore, everyone can understand we all use different kinds of English here. Instead of saying sensor output is x Hz you could have just started with the fact that your probe actually does indeed have this temporal resolution. You specifically brought up the sampling frequency... I wasn't trying to mistake a person for being weak, I'm trying to help them not be mistaken and confusing temporal resolution with something else. I raise people up, I don't bring them. Maybe people aren't like this in your culture and I respect that.


But don't let all this talk about frequency distract you from the fact that you still can't measure turbulence intensity. You measure only a tiny part of the spectrum. You need an empirical correlation to relate 0-100 Hz to tke.