I need help understanding P-V physics.  I built a passive solar greenhouse, with water storage to store heat.  I have gone through several iterations of ways to store the water, but the current arrangement is in large fibreglass tanks.  This is very inefficient at heat capture, and I rigged up a circulation system powered by a 90 Watt, 12 volt P-V panel.  This consists of a little submersible pump which pumps the water up to an old car radiator in the peak of the greenhouse, (where it is hottest).  Two 12 volt computer cooling fans blow the air through the radiator.  And since the only time I will be collecting heat is when it is sunny, a solar panel, (which puts out juice only when the sun shines) is ideal.  So what's the problem?  The fans lasted about 3 minutes, before dying.  They are powered by "Brushless DC motors", rated at 12 volts, 350 ma., that is a little over 4 watts, (driven by a 90 watt panel).  I did not expect that the extra power from the panel would have the slightest effect.  But I think it actually does.  The voltage measured from the panel with minimal power draw is close to 20 volts; as one adds load, the voltage drops progressively.  I am postulating that with these little fans, drawing minimal current, I have simply exposed them to a much higher voltage than they were designed for, and burned them out - does this make sense? 
If my analysis is correct, do I need to place some sort of voltage regulator in the circuit?  Would the units designed to regulate the charge to a battery work, (seems to me, no - that is not what they are designed to do.)  If not, what do I need? (Maybe just 24 volt fans instead of 12 volt ones??)
And, perhaps more significant, am I going to brick my little pumps the same as the fans?

Yes, too much voltage. Break open your bricked fans and I suspect you will find a burny mess.

I would try to find a fan that more closely match the output of the panel.
Many something automotive.

Your pumps might be fine, just cause they are immersed,but then again, maybe not.

Sizing the to total load to match the panel pretty closely might solve your problems.

If you already have the components you want, you could add a resistant heating element to the circut to balance it.

PV panels have "zero load" or "open circuit" voltages above their rated voltage. A fan with a 12 to 24v range should work.
(check the datasheet to be sure.)

You can purchase constant current/constant current controllers for a few bucks. They'll solve your problem, leaving you able to use the current panel with the fans you intended.

These are similar to the ones I use. http://www.ebay.ca/itm/DC-Converter-Constant-Current-Voltage-5-35V-1-3-30V-LED-Driver-Battery-Charger-/370712854629?hash=item5650354065:g:GBQAAOSwP8hTxooy

As you can see, it will cost nearly nothing to resolve your issue. As an added bonus, you'll be able to power other devices from the panel as well. You could, for example, use it as a 5V USB charger as they do in the Nomad GoalZero. You could adapt the DC voltage to pretty much match the output of any wall-wort, so charging 2-way radios, flashlights, battery packs... and so much more... all easy. Note that you'll need one CC/CV controller for each unique voltage, or for each circuit you intend to power. You want to install the CC/CV controller directly before the device you are running, not at the solar panel side.

Go Chris! Thanks for a great answer,one I will probably use myself.

Thanks to Chris.  I had a notion that something like this might be the answer, but was then unsure when I read that they are actually designed to prevent over-charging of batteries connected to a solar panel, and they drop the voltage progressively as the battery reaches charge..

a 12V solar charger will usually have an output for battery charging and another output for appliances.

So it can do these things simultaneously

- take in the 20V or so from your PV panel
- provide a voltage suitable for your battery's state of charge eg 13.2V to trickle charge a full battery upto 14.5V to fast charge a non-full battery
- provide 12V for appliances

You can get charge controllers like this for $10

Meanwhile, put two 12V fans in series, they will get 10V each, which will be fine.

The idea of placing two fans in series is absolutely obvious once one thinks of it, (which I obviously spectacularly failed to do).  Are you willing to educate an old man trying to understand a new technology?  Am I correct in thinking that a 90 watt, 12 volt solar panel, in full sun, is determined to put out as close to 90 watts of power as it can achieve?  If the load on it is only small, (ie. it is drawing only, say 500 ma.), the panel will respond by increasing its voltage, (up to somewhere near 20 volts). (Watts = volt X amps.  Limit the amps, and the only way to compensate is to increase the volts...)  Now, my reasoning would suggest to me that the panel will continue to put out 20 volts up to the point where the current draw (amperage) matches the rated power of 90 watts, in this case 4.5 amps (90 watts/ 20 volts).  At this point, increasing the current draw, (load), will result in a progressive fall in voltage output - to what? the "rated" 12 volts, below which it will not fall any further? And at that point, (a current draw of 7.5 amps), what happens if one tries to draw more than the rated 90 watts of power, (eg. connect a toaster to it!)?  Does it simply remain at 12 volts and 7.5 amps, irrespective of the load? (The significance of the above is that placing the fans in series works as long as the total current draw is less than 4.5 amps.  But above that draw, the voltage will fall progressively, and, as one approaches 7.5 amps, the voltage supplied to each fan will be down around 6 volts, which may well be insufficient to drive a fan rated at 12 volt supply. 
And, what is the effect on the fan of a decreasing voltage?  Does it run slower and slower? If the motor is powering a centrifugal pump does the volume of water pumped per minute simply fall, or does the pressure fall? (or both?)

Oops.  I hit Return to make another paragraph, and it sent my message.  Last question, to go along with the message just sent:  What happens when the solar radiation falls, (the sun goes behind a cloud)?  Do we get the same effect as above, (steady 20 volt output up to some break point, followed by a linear decline in voltage, to  a base voltage of 12 volts and a reduced amperage, proportional to the solar radiation)?

Open circuit voltage (OCV) is panel output when there is no current draw. Actual voltage of your output decreases with panel shading, current draw, and in times of limited sunshine.

OCV is relatively consistent even as sunlight falls off, so you can wire the two fans to get roughly 10V each through much of the day. The problem is that if you increase load to anything substantial, voltage will be unreliable. A load so small as 10% of panel output would cause voltage sway in moderate and low light conditions. In series fans will run, but the speed at which they operate won't be consistent unless you are drawing very little current.

A constant current/constant voltage controller will take variable DC and output a steady supply at your chosen voltage. If panel output falls substantially, you'll still get 12V. Only when the output falls below the threshold (typically 1/2 a volt above the target voltage) will the output fall below your target.

Wiring in series is a good solution if you can't get the proper equipment, or if equipment breaks down and you must restore fan operation immediately. It would be a good fallback position, but I don't think you'd want to use it as your primary design. It's not that wiring in series is a bad idea, but rather that wiring in series provides unreliable results when directly wiring from a variable voltage input.

With respect to your pumps, the precise answer requires a schematic. However, centrifugal pumps typically see a dramatic flow rate reduction with a small decrease in voltage. This is because motor torque is not a linear relationship with respect to input voltage. I suggest giving the pumps the voltage they are rated for.

If you want to better understand how solar panels work and the output you might expect in different conditions, try this link: http://www.wholesalesolar.com/solar-information/solar-panel-efficiency

David Maxwell wrote:Am I correct in thinking that a 90 watt, 12 volt solar panel, in full sun, is determined to put out as close to 90 watts of power as it can achieve?  

Think of it as a 20V battery pack. It can put out 90W (more actually, see below) but only if it is demanded by the appliance's current draw. What you don't draw from a battery just simply isn't provided, but the battery doesn't up the voltage to compensate.

David Maxwell wrote: Now, my reasoning would suggest to me that the panel will continue to put out 20 volts up to the point where the current draw (amperage) matches the rated power of 90 watts, in this case 4.5 amps (90 watts/ 20 volts).  At this point, increasing the current draw, (load), will result in a progressive fall in voltage output - to what? the "rated" 12 volts, below which it will not fall any further? 

The 12V rating indicates that the panel is suitable for a PV system that contains 12V batteries. It assumes that you are running one way diodes, and charge controllers, and that your battery will demand a charging voltage of upto 14.5V. The 90W figure is not what your PV panel puts out, but what the system will put into your battery once the losses from the diodes and controllers are taken into account. Typically a 12V PV panel has 36 cells, each giving an output of 0.55V. Think of it as a battery pack of 36x 0.55V cells in series.

This makes infinitely more sense than my garbled conception.  The critical element here is the concept that a "12 volt" panel doesn't mean it puts out 12 volts.  (I was trying to get my head around the concept of the panel putting out highly variable voltages depending on load.  A chemical reaction (battery) doesn't do this, and it made no sense to me that knocking electrons loose with photons would be any different.)  Sincere thanks to both Chris and Steve for lucid explanations.