Dynamic MPPT algorithm

[blimpyway]

Hi,

Please add your comments regarding the following proposal for an MPPT algorithm.

It's main application would be for windmills or other generators with a relatively high spinning inertia.

Assume we have a controller able to measure how fast the generator spins. For a PMA this should be quite easy by counting the voltage pulses on a single phase.

This controller should also be able to command opening/closing the  circuit to the batteries, so while the circuit is open, the windmill starts to freely accelerate.

While the rotor flywheels, if the generator's voltage pulses are counted every second (or fraction of a second) they are observed as increasing.  For every interval, it can get an acurate measure of the increase of the dynamic energy in the rotor.

Having the rotor's inertia stored in memory (that being measured with other means), the controller can figure out at which interval the rotor had the higher dynamic energy increase, and that RPM is the current maximum power point.

The dynamic energy stored in the rotor varies with the square of its RPM.

A theoretical example:

Before disconnecting the load, the rotor was spinning at 400rpm, computed as being 1600 joules in rotor.

after 1st second:   420 rpm , 1764   joules   (+164  joules)

after 2nd second:   440 rpm,  1936   joules   (+172  joules)

after 3rd second:   459 rpm,  2106.8 joules   (+170.8joules)

So the greatest increase in rotational energy was measured between 420 and 440 rpm, yielding 172 watts.

So for the following minute it connects/disconnects the load acordingly in order to keep the RPM to this measured optimum.

Then it starts again, measuring how the rpm of the rotor varies without a load.

====

1st Note: If the maximum energy increase is in the first interval then this is a sign the maximum power point is below the starting rpm (lower than 400 rpm in the example)

The load must be increased to slow down the turbine to a lower rpm (e.g. 360 rpm) and start measuring from that speed.

2nd Note: this algorithm measures best wind-to-mechanic energy transfer, it should take in consideration the electric efficiency for every given RPM. This hopefully can be done by stationary measuring the generator's output at various speeds and storing this data as a static table within the MPPT controller.

-------

That's all. Opinions?

cezar

[jimovonz]

I think that you may be making it more complicated than is necessary if you goal is limited to obtaining maximum power transfer (may work OK if you have other reasons for calculating rotational energy etc). Measuring the actual energy of the system is not necessary to achieve this goal. In all cases an increase in rpm will result in an increase in rotational energy and vice versa. In your example you can achieve the same result just by using rpm without having to calculate energy.

Strictly speaking I believe that a MPPT algorithm can be implemented based on one measurement - current (into the load/battery). In all cases more current into the battery(load) is better, less is worse. The benefit of what ever you may do to controll the system can be measured by looking at current. If current is going up keep doing what your doing. If current is going down, do the opposite...

It is true that the considerable inertia that our turbines have can make implementing a successful algorithm more difficult. One way of dealing with this is to limit the speed at which we make changes so as to allow the effect of inertia to dissapate and the system to reach equilibrium before we make further changes. This has the dissadvantage of 'missing' optimisation during relatively quick events such as gusts.

I have been playing with an algorithm that uses rpm as well as current as the inputs to a MPPT algorithm. By tracking the rpm we can hold off making some changes when blades are decelerating (when the effects of inertia potentially throw us off) but still respond quickly when they are not.

[Nando]

This algorithm is not practical and is not good to really capture the MPPT energy.

One does not need to know the RPM of the generator, just knowing the input (generator voltage) and the charging current, then the algorithm varies the current by either reducing or increasing the charging voltage, waiting for a while to observe what voltage is generated, if increases or decreases with the current change, to either increase or decrease the charging current.

I have in my User Files : 2IFD_MPPT.PDF which describes an excellent MPPT process done using analog parts to control a PWM Buck charger controller.

Nando

[The Crazy Noob]

Check this topic: http://www.fieldlines.com/story/2006/11/26/22831/362