Part of man rating the Atlas 5 and Delta IV could include delivering water test payloads.
But I wouldn't limit water test payloads to just Atlas 5 and Delta IV and/or a rocket capable of delivering Orion.

Water test payloads should be any size, delivered to specific locations, and orbits that will last more than 5 years.
The amount delivered to one location should around or exceed 100 tons of water. And the amount paid for the deliver of water should be about $500 per lb of water.
And therefore a 100 tons of water at 500 per lb would cost
about 100 million dollars.
One could deliver more than 100 tons to one location or have a couple of different locations [such as at different inclinations]. Or one could pay more for the payloads- one could determine what price one pays by offering to pay 500 per lb, and the lack of interest, could indicate a higher price could be warranted.
But for a test payload and if such a payload had little hassle in terms of requirements and paperwork, I think 500 per lb is a reasonable price and one should in sense "prove without doubt" that such a price is too low, rather than easily and quickly assuming a $1000 or more per lb should the proper price.
So, given that it could be $500- 1000 per lb, and given that more than one location could be selected, and one wants 100 tons of more in one location in a period of time, one could looking at some program with total cost of less than 1/2 billion dollar.

A purpose of such a program is to encourage further testing of rockets. And since in terms of developing new rocket, such testing periods for rockets have a high risk, this type of program would encourage the development of new or improved rockets. If the result of such testing is that fewer govt satellites are lost, this by itself could directly pay the costs of such a problem. But the govt also has an interest in preventing private satellite losses and such as program could lower the insurance costs of rocket payloads and lower launch cost in general.

So, merely putting these water payload into orbit could by itself "pay" for such a program. You could simply throw away or not use these water payloads and it probably be worth the cost of program. But the second part of such a program would be to "buy with" or offer as a prize, the 100 tons of water in orbit to a party that will develop and use this water to make rocket fuel. So, the govt buy 100 tons of water for 100 million, and will in a sense pay 100 million in value in water, if some entity will spend the money necessary to "mine" that water to make rocket fuel. [And of course, they can then use the rocket fuel for whatever purpose to want including selling it to a different party- including the govt].

So, govt would reduce launch costs and possibly get a gas station in orbit for less than 1/2 billion dollars.
The gas station owner, gets 100 tons of water, which if turned into rocket fuel could could worth around $2000 per lb, so having a gross worth of around 400 million dollars and in additional can get more water for around $500 per lb [they offer the same deal as the govt did- or less, if the govt end up paying too much] and can then make 1500 per each lb of water they make into rocket fuel- limited to how much they make per year and how much they can sell per year.

So for a gas station operator, the capital cost might be around 200 to 300 million, and with the capacity to make around 50 million dollars worth of rocket fuel per year- if at $2000 per lb, that's 12.5 tons [25,000 lbs] of rocket fuel per year.
To make 25000 lb of rocket fuel, it would require about 2-3 kw/h per lb, so say 75,000 kw/h per year.
10 kw/h solar array with 12 hours of sunlight per day creates 43,800 kw/h per year.

A communication satellite:
"Measuring 157 feet from tip-to-tip, the satellite's solar arrays feature the latest solar cell technology -- as well as a flat planar design that has proven its reliability aboard the 55 Boeing 601 satellites successfully launched over the past 10 years.

Galaxy IIIC's solar arrays are designed to provide at least 15 kilowatts of power throughout its 15-year design life. The satellite will enable PanAmSat to provide service to the United States and Latin America and will add 77 channels of transmission capability to PanAmSat's geostationary satellite fleet."
http://www.spacedaily.com/news/solarcell-02l.html

So making a solar array which is 15 kw/h or less is certainly possible.
With a gas station one might start small and later add solar arrays. So one might start with an array about 6 to 10 kw/h, and within a year add more arrays, so in a several years you could have as much as say, a 20 kw solar array.