I'm just guessing, but I suspect the EROEI is roughly the same as tar sands: 1.5. That is, for every 2 barrels spent, 3 returned. With tar sands, we get vast environmental destruction that pretty much makes that "gain" a loss. Now ripping up roads, that may be much more productive! We already have the equipment to do it and the environmental damage has already been done. Would the energy gained pay for new concrete roads? Probably not. But hey, there are a lot of roads we won't be needing....

I suspect the EROEI is roughly the same as tar sands: 1.5.

The EROEI of tar sands is much higher than that - this end-is-nigh article estimates 4:1, and this anti-tar-sands article estimates 6:1.

If you want to estimate from first principles, consider that official estimates by Canada's National Energy Board are that natural gas represents the main energy input to tar sands facilities, and that current methods use an average of 0.8Mcf of natural gas per barrel (p.16-17). At 1.05GJ/Mcf for natural gas and 6.1GJ/bbl of oil, we get a ratio of about 7.3:1.

Energy other than natural gas (e.g., diesel for mining equipment) will lower that ratio some, but not particularly moreso than in similar surface-mining operations. Accordingly, estimating from first principles supports an EROEI of about 6:1. Any estimate which is less than half of that is almost certainly due to wishful thinking; I've certainly never seen a shred of support for the claims of sub-2:1 EROEI, despite following the supporting links in, for example, this story. The available data simply doesn't support those claims.