Hi, I am going to replace my roof this summer.  It is constructed with decking over beams with 2.5 inches of rigid insulation, so no cavity.  the only way to insulated it is to add insulation to the exterior.  It seems to me that this would be a good time to add additional insulation.  I'm thinking 3" of poly-iso.  However the house was built in 1950 and has single-pane windows retrofitted with storm windows.  They are a little rattly but the storms seem to do a good job of stopping drafts. The house is located in the Pacific Northwest and is heated with natural gas central forced air unit.  I cannot afford to address both issues so I'm wondering what would give the greatest bennefit for the expense.
Thanks in advance for you input.
Brian

How many square feet of window, and how many square feet of roof? Wood sash or aluminum frame? Is the existing 2.5" insulation rigid fiberglass, EPS or ISO or... ???

Perhaps it's better to think about it this way. The next time you'll get a shot at upgrading the roof insulation will be in 20 years when the moss has taken hold on the roof again, whereas windows can be replaced a room at time or whenever... I don't care WHAT it is, 2.5" of insulation isn't anything like current recommendations for anywhere in the PNW. Whether in the finer analysis the windows might have a better return on investment (which isn't often the case), your golden opportunity on dealing with the roof insulation is right now.

And unless it's a lot of window, I'm thinkin' what you're thinkin'- another R20 on the roof will be the better bet, dollar for dollar. A decent storm raises the performance of pretty-good wood sash window to the U0.4-0.5 range (better than a bottom-of-the line double-pane due to the bigger pane-to-pane gap)- not exactly Energy Star, but not miserable either. If they're not super-tight but still not drafty call it U0.5. The material cost of reducing a 10 square foot window to U0.35 is about the same as adding R20 to 100+ square feet of roof. But the window's heat loss may well be 10x that of the existing roof per square foot, and you may have 1/10th the window area that you have roof area, but the window up only reduces that heat loss by 1/3, while adding the roof insulation cuts the roof heat loss by more than half. Knowing what the actual areas and the actual existing roof R-value will help ball park it better, but rarely do window swaps have any kind of payback on fuel savings unless they've just had it and leak like crazy.

But they can sometimes pay back in creature comfort (especially with a low-E coating)- it's a radiation-temperature thing see:

http://www.buildingscience.com/documents/reports/rr-0307-windows-and-occupant-comfort


The most cost-effective thing you can do to a typical 1950 vintage hot air heating system (if you haven't already done so) is to seal all seams & joints with mastic, caulk the interfaces at the registers, and FSK tape (2" aluminumum tape) the panels to the furnace to minimize leakage. Most homes won't take more than $50-100 of material (and a lot of patience), but it typically returns high single-digit, and sometimes double digit reductions in fuel costs (particularly when the ducts & furnace are in air-leaky semi-conditioned basements or crawlspaces.) Even if for some reason it buys you very little in fuel economy, it'll be quieter.

Thanks for the thoughtful reply.  you have confirmed my suspicions about what my strategy should be.  Insulation is not as sexy as new windows but it does seem to make the most sense.  I think the existing insulation is rigid fiberglass.  Our existing windows are well shaded on the south and there are no windows on the west.  we do have 4 very large windows 35sf ea. on the north but they are fixed w/ good storms. overall they seem to be fine for now.....next i need to spend time in the 16" high crawl sealing duct-work. yee haa!

  Enjoy! (If you can call it that...)  I don't s'pose your crawlspace ducts are insulated either, eh?

If  the crawl space has open vents (often do), converting it into an insulated sealed mini-basement (micro basement?) with a poly vapor retarder floor and XPS insulated walls can be a pretty big efficiency boost too- could be 25% or more of your heat loss is out the bottom (after you insulate the roof, that is...).  The fact that your ducts live there may make it worthwhile. For some discussion on the subject, download this:

http://www.buildingscience.com/documents/reports/rr-0401-conditioned-crawl-space-construction-performance-and-codes

There is more on the subject on the BuildingScience site as well.  material costs would be low, but the pain-in-butt factor is likely quite high...  (Having spent time insulating a 30" crawlspace  I suspect insulating & sealing it from the outside might be a better method.)

A decent not-too-tough tool for figuring out what the next-most cost effective hack would be, try filling out the 101 question profile on your place here:

http://hes.lbl.gov/

(Save the session number- they'll store your profile data forever, so you can do it a piece at a time, modify it as you figure out more, etc.)

It's based on local weather data, and is a recursive model calculating annual heating/cooling loads & gains on a more continuous basis (it'll factor in average cloud cover, not just temperature when figureing solar gains, etc.)  When you get the entire profile filled in an run the calc, check the energy use numbers against historical reality- if it's within 25% of reality, it's close enough to work with the cost/benefit analyis "upgraded details" spreadsheet it generates. If it's off by quite a bit, figure out what might be wrong/tweakable in your profile to get a better fit, then run it again.  Be sure to insert your ACTUAL delivered-price for utilities in your profile rather than accept their database numbers for the "state average", which can affect the cost-effectiveness numbers dramatically.

The cost-benefit stuff has crazy numbers filled in for actual costs- do your own estimates (or get quotes), fill real numbers into the cost-benefit spreadsheet an have it readjust.   Generally anything with a sub3 year ROI is a no-brainer (like duct-sealing), but the top 5 things recommended (once realistic guesstimates are inserted) usually deserve a closer look with a sharp pencil/calculator. It definitely changed the order in which I've been working retrofits on my place. (And better windows are currently down at number 6 or 7 on my modified list, and a high-ticket hack at that.)