Measurement-foo and shelf-fill rate help?

November 30, 2004 9:03 AM

See a librarian’s brain asplode*! So my library is moving back into our building next summer. We have some 8,400 linear feet of bound serials* being moved into 15,500 linear feet of compact shelving*. We need to tell the mover what we want our fill rate* to be. Sounds easy, right? However, there are complications, horrible complications.

Lingo:

Asplode – explode painfully
Serials – mainly journals that have been collected and bound together into volumes
Compact shelving – moveable shelving that slides to open or close aisles as needed
Fill rate – the percentage of the shelf that is filled by books. Ideally, you don’t want to go over 80% because it makes shelving new items difficult.

Other factors:

Each serial shelf unit is five shelves in height. Each shelf has 11.5 inches between it and the one above it.

Each shelf is roughly three feet wide.

The complications:

Some volumes are either too tall or too deep for the shelf. Tall items require that a shelf be pulled from the shelf unit to increase the 11.5 inch clearance. This will decrease the number of shelves in the shelf unit. Some items are too long to fit on the shelf, so when the compact shelving is closed, they would jam against the opposite shelf. To make space, a shelf must be pulled from the shelf unit behind the one with the long item in order to keep the spines flush. Some volumes are both too long and too tall for the standard shelf unit.

For the sake of this puzzle, let’s assume that we can not create a separate section for serials who are too large for the shelf. While my library does have a Folio section that would likely take all the items that are both too tall and too long, it is in the best interests of accessibility to keep all the serials together. Besides, these volumes are collections of bound issues of a journal title and must be kept together. However, one thing journal publishers love to do almost as much as change their journal’s titles is change the size of their journals. So a serial title could have 20 volumes, of which only 1 is oversized. We can’t pull that lone volume out, and we can’t pull out the entire title.

There are 405 linear feet of oversized items in the collection. 23 feet of them are too long, 290 feet worth are too tall, and 92 feet are both too long and too tall. They are not necessarily all next to each other, although some are. Each single oversized item can cost anywhere from 3 to 9 extra feet worth of shelf space as shelves above, behind, or above and behind it are removed. I have an inventory of all the oversized items.

This collection grows roughly 400-450 feet per year. The oversized elements of this collection grow only 14 feet per year. We are hoping for 10 years minimum worth of growth space with an ideal of 15-20.

If we give a fill rate that is too low, causing there to be books left but no shelf space, days are lost as we shift and the entire move schedule is out of whack. If we give a fill rate that is too high, we cost ourselves growth space that has to be justified to the school. Either way, it’s a “Very Bad Thing.”

I figure I need to produce a pretty detailed shelf map of what the stacks should look like once the movers finish. However, outside of trial and error, I have no idea how to produce this map.

The Questions (finally!):

Does a computer program exist that can perform this mapping for me? If not, how hard would it be to code said program?

If computers can not help us, what other method would you recommend for producing this sort of map/plan?

Am I doomed and should I start eating bugs now?

posted by robocop is bleeding to grab bag (15 comments total)

"constraint programming" is what you're looking for. there may be packages out there, but you can also get dedicated languages. one nice example is mozart/oz. try reading this.

however, solving this problem from scratch may be painful. hopefully there's either a library specific solution already out there with a nice easy interface, or you're at a university or similar and can go talk to the computing dept.
posted by andrew cooke at 9:11 AM on November 30, 2004

Graph paper is probably the best solution.
posted by smackfu at 9:33 AM on November 30, 2004

It sounds analogous to furniture rearranging, which I've heard always works best with scale cutouts of the rooms and the furniture in question.
posted by Alt F4 at 9:51 AM on November 30, 2004

But I hope there's an easier solution. Sorry to not have better info for you.
posted by Alt F4 at 9:52 AM on November 30, 2004

Personally, I'd draw all the shelves to scale in Autocad, define the space necessary for each journal, move eveyrthing around, label the whole mess and plot it out.
That's just me, though.
posted by signal at 10:26 AM on November 30, 2004

No computer program, unfortunately, but maybe a few ideas...

Will the serials be sorted by call number? LC? If so, do you have breakdowns of linear feet and annual growth per class (A's, B's, etc.), accounting for both the regular sized and oversized volumes? It sounds like you have the numbers for the oversized material anyway.. If you do have the numbers, do as smackfu and Alt F4 suggest and bust out the graph paper and/or representative scale cutouts.

If you don't or can't get numbers to such a granular degree, then do some guestimates, think about consistently leaving the top or bottom (or both) shelf of each unit empty for future growth and shifting and err on the side of caution when it comes to the serials you tend to collect heavily in. For example, for a typical research library, that might mean leaving extra extra space for those Q's and Z's.

Oh, and keep in mind that all the planning in the world might not save you from some trial and error work by the end. At my last library, we had library movers who were a little imperfect and wound up doing a lot of shelf-reading after they were done (but thankfully not too much additional shifting.)

And this wasn't part of your puzzle, but if you can separately shelve those items that are both too long and too tall, it seems like that might help you save some serious heartache in the long run-- i.e. storage projects, etc. that might require future major shifts. And would your library be okay with shelving the items that are just too long spine-down so they can be handled the same way as the items that are too tall?

Good luck!
posted by bricsot at 10:32 AM on November 30, 2004

Oh, God, we're just about to jump headfirst into this shifting/renovation hell, ourselves. I may contact you, robocop is bleeding, to see how things are going...

I don't have any suggestions, other than to leave every top shelf empty in the new space, and (if you can) leave an entire range empty every few rows or so.
posted by arco at 10:44 AM on November 30, 2004

Hi all- Thanks for your answers so far. I've been muttering about this to myself all day, wandering through the temporary stacks area with a tape measure and a lost expression on my face.

Our serials are shelved based on a seven digit code based on the first two letters of each word in the title. So it's a rough alphabetical.

I have a pair of minions I can sic on measuring and inventories. At the moment, I'm considering doing a peaks and valleys measurement of the stacks using my oversized item inventory as the peaks with the valleys being all those good little books that fit nicely on the shelves. I could then take up the graph paper and try to figure things out, but consider this...

I take a long sheet of graph paper and break it down into shelf units and shelves representing the first row in the serials stacks. It is two sided, just like the compact shelves. I start with 0000000 and work my way down the row. For each oversized by height item, I remove a shelf from the shelf unit. For each oversized by length, I remove the shelf behind. For an item that is both, I do both. I work my way through the first side of the first row, then turn the corner at the end and begin on the reverse side. I continue until I get an item that is too long by depth. Normally, I would remove the shelf on the opposite side of the row, but I've already plotted that portion out! Gah! Changing the location of items previous would affect the locations of what I am working on.

arco, the best advice I can give you is to get an accurate shelf measurement right now. Ask about to see if anyone knows how many linear feet your collection is. If they don't know, start measuring. If they do, ask where they got that info from. If they answer with anything besides "I measured last year," start measuring. Our library based a lot of planning on inaccurate shelf measurements and while it was not as bad as it could have been (we thought we had more than we did), it did cost us some project time once I found out! For other advice/mea culpas, feel free to drop me a line. If you're going to be at ALA in January, we can weep into a beer or five, too.
posted by robocop is bleeding at 11:07 AM on November 30, 2004

Are the compact shelving units double sided? /cringing architect
posted by Dick Paris at 12:07 PM on November 30, 2004

Dick Paris, ayup. Books can be placed spine out on either side of the compact shelving row so that they can be opened or closed to create/conceal new aisles.

This is great for stuffing more books into a set amount of space, crushing pennies and patrons under the shelf tracks, and pretending to work in a big bank vault when you turn the wheel to open the aisle.
posted by robocop is bleeding at 12:25 PM on November 30, 2004

Pure conjecture/not knowing how things actually happen: Won't the books have to be in some sort of order (aside from "oversize" materials)? If they're supposed to be ordered aren't your options for arrangement fairly limited? How much flexibility in storage do you actually have, aside from separating out the oversize books from the good little obedient books? That's at least what my school's architecture library did, since pretty much any architectural monograph will be oversize. If you looked up a book in the catalog, it would either be in "regular" stacks or "oversize", so there were really two separate categorization systems.

But damn, that's a more than a mile o' books. Eep.
posted by LionIndex at 12:31 PM on November 30, 2004

My first instinct would be to find someone that has done this before. There must be some kind of librarian communities and resources online. This one recommends some books that might help. Maybe you even have them packed up somewhere. Why reinvent the wheel?

I'm thinking that the growth rate depends on the the letter of the alphabet. So leave more room for Ss than Xs.
posted by euphorb at 11:23 PM on November 30, 2004

I believe I can offer some suggestions. I work in the Collections Maintainence department for the Penn State University Libraries. My explanation could be kind of long so I might post the first part and you can respond and tell me if you want to hear more, or if you already know what I am talking about. I may switch back and forth between feet and inches as you use feet and we deal almost entirely in inches (for reasons that will become apparent later) We use a technique called the geometric mean for using almost all of our collections. This can be broken down into steps the first of which is:

1) Make a separation on paper of the collection into smaller discrete units. So in your case perhaps A, B, C etc. Measure and map each of these discrete units. For this technique to work you are going to need to know how many inches of each unit there are and once you start moving how many inches of the unit you have left to move. The easiest way to keep track of how much you have left to move is to have a map with numbered sections on it corresponding to a spreadsheet that has the measurements for each numbered section.

2) Determine the future space requirements of each unit for the period of time in the future that you are planning for. The roughest approximation would be to assume that future growth is going to be exactly the same as past growth. In that case (# of feet of As)/8400 is the percentage of the current collection that is As then multiply by 14000 to get the future amount of space at capacity that should be allocated for the As. This is going to provide a pretty rough approximation. An alternate method would be to measure the last full year (2002 or 2003 depending on where you are in the binding process) of each journal set for each unit then add them all together. Then the space you allocate to As (in inches)would be (Current Inches of As) + (Growth of all journal sets in As in inches * # of years you would like to plan for) you may have to juggle the number of years you are planning for as you realize that you either have a lot of space left over when you get the end of the alphabet or you don't have enough space left over. Either way the number of years you use should be constant each time you run through the numbers. IE run the numbers for units A though Z using 10 years, then 15 years then maybe 13 years.

This method also has the benefit of giving you some idea of how long you will have before you run out of space. Of course, you don't really have that much time but in the library business that's the way it goes.

Well, this was the first part. If you think it might be useful just post some kind of response and I'll go ahead and post some more.
posted by jefeweiss at 6:24 AM on December 2, 2004

I just reread my post and I've got to say, I shouldn't ever write anything before noon. I'm not very good at using commas at the best of times, but before noon they are apparently completely optional.

Here's a few to sprinkle into my last post.

,,,,,,,,,,,,,

As far as terminology we refer to bound periodicals as journals instead of serials, but you probably figured that out.

I forgot to mention the advantages and disadvantages of using the geomean method. Using the geomean you shift say ten sections, then you do some fairly simple calculations, then you shift ten more sections.

On the plus side, you don't have to plan out where everything is going to go ahead of time. It will adjust itself as you go along, so if you lose a shelf because you have an oversize book the next time you recalculate your load rate it will take that into account.

On the minus side, you have to recalculate your load rate every ten or twenty sections. This can be a problem if you use contract movers, because you have to have someone oversee the work. In my opinion, this is probably a good thing as the last time we used contract movers they screwed a bunch of things up.

Also, if you use just the plain geomean method you load at a (between each calculation) constant rate. So you wouldn't be compacting journal sets. If you have a lot of short journal sets it doesn't really make too much difference, but if you have some really long journal sets it can be a pain in the rear. There are also ways to fudge the geomean so that you can compact journal sets, but I'm not going to go into that right now.
posted by jefeweiss at 8:59 AM on December 2, 2004

Thanks, jefeweiss! Our terminology for journals/serials/etc is pretty wacky. A title is a journal when it is unbound, but becomes a serial when it gets bound into a volume. I think this stems from an ancient line of demarcation between some of the tech services folks.

In a meeting yesterday, the decision was made to pretty much ignore the oversized by depth books. So the problem of a long book messing up the sheves of the items behind is no longer an issue. This frees up mucho sanity. It also means I can unleash my minions to measure the collection, taking into account those oversized by height, and plot out the collection pretty easily.

This is a major load off, but now there's wacky office politics messing around with things, so I'm pretty much zero sum in regards to asploding.
posted by robocop is bleeding at 8:09 AM on December 4, 2004