Fortsätt till huvudinnehåll


Visar inlägg från 2014

[VAWT] First practical tests

The first practical tests have commenced! The windmill below is built within this Bachelors Thesis: 

We had worked in small subgroups, where I had focused on calculations, software and electronics. Some other guys had done the mechanical design and therefore I didnt't see the turbine first until the day all parts were to be assembled. What I hadn't realised before is that the turbine is huge! It measures some above 3 meter in height. Here is a rather crappy first movie of over first test run:

 Turbine control is done via electronic switching by using PWM. A Maximum Power Point tracking algorithm is tried here. Among other things wind and electrical output (voltage and current) are measured that can be used to evaluate system performance. This can be used for estimating turbine output.

The current measurement was bit of a challenge to design as it had to work over the unusually large range 0,5- 60 amps. The measurement is split up in to ranges in hardware in this design. Here i…

[VAWT] Relation between rotor radius and conduction losses

Since the last post a lot of effort has been dedicated to a presentation and mid-project report so little concrete work has been done in the project itself.

Today I however played around with some equations and got a new one of interest. I haven't double checked it yet though so full disclaimer. However, it seems that the steady state conduction losses in relation to it's produced shaft power in a DC machine connected to a Savonius turbine with a radius r can be expressed as

\$ P_{cond.loss,frac}=\frac{P_{cond.loss}}{P_{shaft}} = \frac{R \rho H V C_{p0}}{\lambda_0 k n}r^3 \$ (1)
where V is the wind speed, R is the DC machine series resistance, Roh is the air density H is the rotor height, Cp0 denotes the turbine effiency and lambda a desired tip-to-wind-speed-ratio, both assumed to be constant (a good approximation if a controller is used). Furthermore k is the ideal DC machine constant and n is a gearing ratio between the turbine and DC machine shaft assumed to have no losses…

[VAWT] Second test

There were some issues during the first tests due to that the output voltage from the generator was too low. The generator voltage can be raised by using a gearing factor n that increases the rotational speed and therefore the EMF voltage by 
EMF = n*k*w
where k is the motors voltage constant and w is the wind turbine's rotational speed. The generator will spin faster and therefore generate more voltage, simply put. 

When a gearing is introduced the torque on the generator's rotor is also decreased. This reduces the copper losses 
Ploss,copper = r*I^2 ~ r*(T/(k*n))^2 
as the current I is directly proportional to the torque T, and inversely proportional to a theoretical motor constant k, in theory. Lower rotor torque thus means that the copper losses are reduced. 
Reluctance effects However, the BLDC motor used is affected by quite a lot of  reluctance effects. An initial torque of a certain level is needed to overcome the reluctance effects and the motor then rotates a step. Th…

[VAWT] Plausible improvements of the Savonius windmill

Cut a barrel in half, mount it on a shaft and you have a windturbine. If the barrel is used, then what else probably would end up on a scrap heap is instead used for converting renewable energy.

Simplicity and low cost makes this an attractive option especially for societies with limited economy and a malfunctioning or non existing electric grid. Small off-grid electrical networks can be built and people who perhaps most needs electricity get that. Isn't that neat?

Savonius (from

An overlooked potential?
The Savonius though has a widespread reputation of having low effiency and is often dismissed as a credible option around forums and in formal litterature. However, when looking at the graph below from a publication Wortman did 1983, the effiency can be realtively high provided that the TSR(Tip-to-Wind-Speed-Ratio) is held at a correct value and the windmill should work quite nicely. In practice this could probably be done by controlling the generato…