The energy savings will never come close to 6%. First, the intention of voltage reduction is to reduce demand during peak times not reduce energy sales. Unfortunately, the first hour is were you see the most savings in demand. The reason is the loss of load diversity. Think of it this way. Since it will take longer to raise the water temperature in a hot water heater, an oven, a room that is heated by a heating element, or a dryer, the number of resistive loads will gradually increase and you will lose quite a bit of load diversity. Resistive lighting is the only load that you will benefit throughout the voltage reduction period. As you know, most commercial lighting is fluorescent, HID or other energy efficient technologies that use a ballast or driver that regulates the lamp voltage (constant power device) within a input voltage range of +/- 10%. There will be little to no savings for lighting. This is the same for most inductive loads. Therefore, only the resistive loads will generate demand reductions and the first hour is the most productive. By the way, I worked for a Cooperative that has been doing voltage reduction for over 30 years. If you have know what you are doing, it is a very effective way of reducing demand. Also, VAr support is not the most cost effective solution. The best solution is to design the distribution system for voltage reduction during critical demand periods without low voltage problems at the end of the feeder or tap. In my opinion, every utility should be increasing there distibution voltage to 24,940/14,400 V. Losses are less, voltage reduction can be accomplished more efficiently, ground faults are easier to detect, etc. This is one thing that the Federal government (REA now RUS) did correctly.