Monday, November 17, 2008

Saving the world, one solar panel at a time

If you saw my previous post you probably thought "wtf is this?". I'm sorry that I did such a poor job of labeling anything on the graph, but I was really excited and it was sort of a rushed post.

Anyhow, as all of you know I've been stressing out over my research project because I still hadn't gotten any solid results worth publishing (deadline is less than 2 weeks). But now we did! and I want to share the loooooooooove :-) Get ready for me finally getting technical.

How solar panels work
Your typical photovoltaic (PV) cell is made out of silicon, though there are many different types and technologies that are quickly gaining better reputation. In lamest terms (and because I actually don't know much more detail about it) a photon is capable of kicking off an electron from the silicon valence band, effectively creating an electron-hole pair that flows creating electric current (and power!). Sounds very simple but not all photons are capable of creating the e-h pair. For this to occur the photon must have at least the same amount of energy as silicon's bandgap energy. If the photon is higher in energy the excess transforms into heat, same thing with low energy photons, they don't produce e-h pairs but they are still absorbed by the material. This is the reason why solar panels get hot, very very hot.

The solar resource
In case you did not now, solar panels are extremely sensitive to shading. Shading from clouds, neighboring building, trees, birds, etc. The relation between solar irradiance and output power is linear, however the relation between shading and output current/voltage is not. We only care about this if we put several solar panels in series or parallel to form a system, because then all the panels interact with each other and the operating point of Panel 1 will heavily depend on the operating points of Panels 2, 3, 4... n.

The series string problem
Solar panels produce direct current (dc) power, while pretty much every appliance in the world uses alternating current (ac) power. In order to interface the two you need an inverter. Types, sizes, cost, efficiency of inverters varies widely depending on the application, but one thing is known: the smaller the inverter the higher its efficiency is. Inverters tend to be big because more often than not they need a large transformer to convert the output voltage to 120V-ac, and transformers just like any type of magnetics tend to be lossy.

Typically, in a residential or commercial installation you would like to build up the dc voltage of your system and maintain it relatively constant in order to maximize the efficiency of your inverter. Hence everybody would like to put their solar panels in series, but this brings on another challenge as well. If the panels are in series they all share the string current, which means that the panel with the lowest output current will dictate where the rest of the system operates. You can bet all your money that the PV panels will operate far from its maximum rated power, and your entire system efficiency goes to hell. A single cell out of hundreds in the system can cause malfunction of the entire string.

Module integrated converters (MIC)
To solve this problem many people have proposed inserting a high efficiency dc-dc converter at the output terminals of each solar panel before interfacing to the string. Each converter performs what is called Maximum Power Point Tracking (MPPT) to ensure that the panel is operating at the point of maximum rated power. This way each panel is effectively decoupled from the string and it is insensitive to any changes in the system. This solution has been proven many times before BY SIMULATION. So far there hasn't been any published research on a real system with actual hardware.... until now... And let me tell you why there isn't published research on real systems, BECAUSE BUILDING THIS STUFF IS HARD!@%#~!!!

Last Saturday Sara (from the building systems group in the Civil Engineering department), Scott (undergraduate EE working for my advisor) and I went outside for what I thought would be another failed experiment. But luckily I was wrong and the experiment went really, really well. What the graph from the previous post shows are performance results from the three cases we studied. The blue bars represent the series string without any converters, three solar panels in series, period. The green bars are the system with my converters and the red bars are the percentage power gained by integrating the converters in each case. The first set of three bars is when all three panels were at full sun, you can see the gain is negative because there is a tiny bit of power used by my converters, but this is still a kick ass efficiency. The next set of bars are the system under "medium" shading conditions. I can't tell you what "medium" is because we haven't defined it yet, but it implies significantly lower solar iiradiance. The important thing is that output power was improved by nearly 20%. In the last case, the system was under "heavy" shading conditions, and the converters helped increased the output power by almost 40%!!!! These are really excellent numbers and I was very pleased with them.

It has been 6 long months since I first started building the circuit. Some days were longer than others, but I'm glad that now I get to relax a little bit more. Not just that but when good things happen I tend to feel more motivated to work. This research will be published next year in the Applied Power Electronics Conference 2009 happening in Washington D. C. next February where yours truly will be presenting this work.

The day after making history
As you expected I went out last night to celebrate this moment of history-in-the-making. I drank waaaay too much, and to make matters worse the bar we went to had karaoke. It was pure anarchy, by the end of the night I wasn't seeing double, but triple. I was hugging strangers and talking to them about solar power, it was ridiculous and wonderful. Most of today was spent in bed feeling sick, *sigh*. This is going to be a great week.

Thursday, November 6, 2008

A comparative study of two useful discrete-valued random fields for the statistical modeling of images*

*Is the title for a Computer Science paper from the late 80's. No I didn't read it, just borrowed the title.

I did this yesterday:


Which is a huge improvement from this:



When I am done with what I'm doing right now, I will share it with you.

Some free publicity:
My friend Dan Hill is a "modern day prophet denouncing the hypocrisies of our time". He is also a very talented artist, writer, filmmaker, historian, librarian, entomologist, space cowboy, time-traveler and connoisseur of all that is vice. He has a bi-monthly periodical that exposes, well... the hypocrisies of our time. Each issue is a journey through the deepest darkest secrets of human nature.

You can order it via his website "The Fifty Flip Experiment" for $2 an issue, or buy a 1-year subscription for $10 (includes shipping!).

Me and Dan once drew humanoid elephants together.

I hope he doesn't sue me for copyright infringement.