I would use process sight glasses at evenly spaced intervals up the tank shell. You could specify the materials of construction to suit the service, and they are intended for this purpose. I do not think it necessary to know the exact level, if the sight glass interval is small enough. However, an ultrasonic or radar level indicator would likely be less expensive.

Thanks for the suggestion fegenbush! I had considered this, but I think the risk of freezing goes up and it would become difficult to insulate. It also would only give you a sample at the intervals and not a true visual on the full range of the tank. It also wouldn't necessarily show you how much gunk (solids, wrenches, lunch boxes, etc) is sitting at the bottom of the tank because this stuff would probably plug up a sight glass. I'm also thiking it may be a bit costly.

I like the window idea because I was also thinking I could mount my oil skimmer in front of this "window" and then I could still see all of the above, and also see how immersed my oil skimmer is in the oil layer and get an accurate inventory of oil in my tank. Then I wouldn't have to guess. I'm thinking other industries may have had need to develop a tank like this.

A cylindrical tank is very efficient at containing liquid pressure because the pressure loads the shell in hoop stress. Slotting the shell from end to end just cuts the most critical part of the tank in exactly the wrong way.

The stress distribution across the rest of the tank would still presumably be relatively evenly distributed in the hoop direction, and yes as you approached the point of this slot you would create stress concentrators wherever you needed to bolt which would need some thought which might include beefier connections at this point. Presumably, this would not be that dissimilar from that of a bolted tank design?

If the window material did not come with adequate mechanical strength to handle the hoop stress, equally spaced bars could be attached from one side of the steel shell to the other to nullify most of the requirement the window would have in helping hold back most of the hoop stress. Guess you'd still need to consider radial stress though. I was thinking the window could be put on the inside of these bars with a sealing surface between the tank shell and the window and then have the pressure inside the tank push the window into the metal sealing surface.

Just some thoughts I guess...

I'd question how long you'd be able to determine the interface. Level sight glasses in dirty services are notorious for fouling and being close to useless.