How to modify the filter to match the power output of the wind turbine with the wind speed?

hello,

I have the wind turbine power results on the left and the wind speed on the right, both varying over a 24-hour period. From three different scenarios, it's clear that the power output doesn't align with the wind speed. Could this be due to the filter I uploaded? If so, how can I modify it?

thank you in advance

Answer 1

[William Rose]

@jana nassereddine,

The horizontal axis labels and values are partly cut off, or absent, on the three images you uploaded which contain two plots each. It appears that the horizontal axis labels for the left side plots are "Load (kW)". If that is correct, then the plots are not power versus time. Do the right side plots (with black background) show wind speed versus time? Please plot power versus time and wind speed versus time, with visible labels and units and values, for both axes. Then we will know what is being compared.

The system diagram shows a sampling rate of 1e-6, but the right side plots appear to be sampled much more slowly. Please figure out why this is so, and why the sampling rate appears different for the right and left side plots.

I don't think the lowpass filter is the cause of the mismatch between wind speed and power output, because the power waveform in the plots does not appear to be a lowpassed version of the speed waveform. I suspect the difference is due to the inner workings of the wind turbine power system box, or the difference is because you are not plotting what you mean to be plotting.

[jana nassereddine]

Here’s the figure again, now with titles on the x and y axes. Regarding the sample time, I’m unsure what the issue is. The sample time for the microgrid model that includes this wind turbine is set to 1e-6, while the sample time for the MPC is 0.01. However, this shouldn't have any impact, right?

PS: I included the load figure to illustrate the x-axis for the wind power figure.

[jana nassereddine]

I adjusted the filter parameters for the sample time and constant time to match those of the AC power wind output. Now, for the same wind speed, you can see the updated wind power in kW. I also incorporated the AC power from the wind for you to see.

[jana nassereddine]

I adjusted the sample time from zero to 1e-6, but there hasn't been much change.

[jana nassereddine]

[William Rose]

@jana nassereddine,

Thank you for the additional plots.

Did you edit your original post? I am asking because I can no longer see the system diagram, which you had attache as a figure in your original post.

I agree that a sampling time of 1e-3 versus 1e-6 does not make a difference. But I am still puzzled by the fact that the wind speed scope signal seems to be peicewise linear with corners about every 0.08 seconds. It looks like a signal that is sampled at about 12 Hz. But let's ignore that for now. Let's focus on the plots in your two most recent comments.

I am not a simulink person. Therefore I don't understand the difference between the white background plot labelled "Wind Power (kW)" and the black background plot labelled "Wind Power AC (Watts)". The former is > 10 kW and the later is <3 kW. Why? It's not due to a lowpass filter.

In your two most recent comments, the first order lowpass filter time constant is 0.015 s. The difference between the wind power waveforms and the windspeed waveform , in the two most recent comments, is not due to the first order lowpass filter. Wind power is roughly proportional to velocity squared, i.e. a nonlinear relationship. A real wind turbine has complicated dynamics and nonlinearities. I don't know what lives inside your wind turbine block, but I assume it has some of the complictaed dynamics and nonlinearities of a real wind turbine. Therefore it is likely that you should not be surprsed that the wind power waveforms do not have a simple relationship to the wind speed. There is probably no simple linear filter that could make the wind power waveform look like a scaled version of the wind speed waveform.

A person who is competent in Simulink could provide better advice than me.

[jana nassereddine]

thank you for your reply;

Now, about the data sampling. You've got 24 data points over 2 seconds, which means each data point represents a time interval of 0.083 seconds.

I used an AC/DC converter to turn the wind's AC power into DC power. That's why I have two power sources – one AC and one DC. The former is <3 kW and the later is > 10 kW.

[jana nassereddine]

this filter is for the first plot

this second filter for for AC power the second plot;

[jana nassereddine]

My question is: why does changing the wind speed data, like in this case, result in a delay between the wind speed and the power output? As you can see, it takes time for the power to increase.