http://www.irrigation.org/ibt/0208/p29.htm
The CSV article has much more clout because there's actually math used.
"If a variable-speed drive had been used the submersible motor would overheat or "cycle" during low-flow situations."
-Perhaps in 1992. Nowadays, decent vfd systems automatically detect flow and shut off accordingly. You can cycle a pump as many times as you want via a VFD, there is no inrush current and therefore no heat build-up.
Let me take this one thing at a time. The article in question was talking about a 75 HP submersible in 10" casing. A CSV would derate the motor to about 40 HP at low flow. Derating the motor would allow it to safely pump hot water, so it takes very little cool well water to maintain proper cooling. This is why the flow can safely be as low as 5 GPM with the CSV.
A VFD does not derate a motor, it creates a smaller motor (40 HP) from the larger one (75 HP). Cooling flow still needs to be adequate for a fully loaded 40 HP motor, which is .5 feet per second. With an 8" motor in 10" casing that would be a minimum flow of 55 GPM with the VFD. When the flow needed is less than 55 GPM, the VFD would sense low flow and shut off the pump. Flow is still being used so the pressure would drop and the pump would be started again. This cycling would continue as long as the flow needed was less than 55 GPM.
Seven years ago when that article was written, there was no such thing as "sleep mode" for a VFD. Because of that article and the many other times I said it, "sleep mode" was added to most VFD controls. When a VFD shuts the pump off because of flow being less than 55 GPM, "sleep mode" allows the pressure to drop, usually about 10 PSI, before the motor restarts. This allows a pressure bandwidth, so a pressure tank can supply some water during times when the pump is off. Before "sleep mode" the pump would have restarted almost immediately with a VFD control, which would have caused excessive cycling and heat. "Sleep mode" gives a little more time between cycles but, you would need to use a very large pressure tank to keep the pump from cycling excessively.
A submersible must reach 30 hertz or ½ full speed in 1 second to prevent thrust bearing damage. You can't get from 0 to 30 hertz in 1 second without inrush current. This current would be less than if the pump went to full speed. However, inrush current will still be 200% to 300% of running amps with a VFD, compared to 600% with across the line controls. So I disagree that a VFD will let you cycle a pump as many times as you want. It is usually the fact that the motor has not had time to cool down before it is restarted that is the biggest problem.
There is not quite as much heat build up from starting with a VFD as there is when starting across the line. However, once the pump is up and running, the VFD causes much more heat and is much harder on the motor than starting across the line. The harmonic currents created by the VFD cause additional motor heat. Even with required filters on the VFD, voltage spikes to the motor can be 400% of line voltage. With 480V, the VFD sends spikes to the motor of about 2,000 volts. This adds to the motor heat, and also causes partial discharges in the motor windings, that eats away and breaks down the insulation. These are the main reasons why a VFD controlled motor needs more cooling flow than a motor running on standard across the line controls.
Now we can also talk about resonance vibration, harmonics that are fed back into the electric grid, bearing currents, and many other things that are problems when using a VFD. I am presently working with an engineer who swears that VFD's are causing the steel casing in his city wells to deteriorate. An example is two wells drilled at the same time, using the same casing, which are now 9 years old. One well used a VFD controlled pump, the other used standard across the line controls. The well with the VFD has had to have the casing replaced as it looked like Swiss Cheese. The casing in the other well still looks like new. VFD's are not the cure all that everyone wants them to be, and in most cases, VFD's cause more problems than they are worth.