OK, here is the deal (as I understand it)... These kind of figures are always approximations at best. But it boils down to something like this:
The formula for BTU/hr is as follows:
BTU/hr = (watts * 3600) / 1054
so, (1300 * 3600) / 1054 = 4440 BTU/hr.
The reason that Cisco lists a higher value is that there is a certain inefficiency to converting AC to DC (i.e. the supply is rated to PROVIDE up to 1300 watts but will also CONSUME some number of watts in the conversion process). It is common for the efficiency to be anywhere from 60 - 85%. In this case it is apparently 72%.
From this, you can see that the BTU/hr figure is actually worst case if the power supply were being stressed to provide the full 1300 watts of RATED power. In order to calculate actual power consumption, you have to know A LOT about all of the internal circuitry in the chassis (you really don't want to get into resistive vs. reactive power, etc. unless you are pursuing EE).
Having said all that, there actually is a page that lists the DC current consumed by each module (again, I'm sure it is worst-case). So you can do some math (P = I * E) to try to estimate how much of the rated 1300 watts are actually being consumed by the modules. Make some assumptions about inefficiencies in the chassis getting the current to the modules. Then redo my calculation above with the total estimated value.
Finally, as for redundant power supplies, things are a little murky here as well. It is reasonable to assume that there is some type of load balancing between the two supplies. However, it is also reasonable to assume that no more than 1300 watts will be drawn (total) from both supplies together, since one is capable of doing the job by itself. Now it is debatable as to whether you need to consider the efficiency issue twice since AC is being converted to DC twice. I suspect that there is some minimal impact to BTU/hr, but probably not worth considering - especially since 6140 is worst-case and unlikely to ever be realized.
So, in closing, again all of this being as I understand it (never finished that EE), I think 6140 is a reasonable estimation regardless of whether you use redundant supplies and regardless of how the chassis is populated. It is somewhat of an "accepted practice" to just use the manufacturer BTU/hr rating for planning purposes.
Scott
