One suggestion I have is that if all the work stations are all using the same image and systems are alike, swap the system that is frequently having this problem with one that isnt and monitor to see if the problem moves or remains.
If the problem moves then you know its something wrong with that specific workstation, if the problem remains then its something on the network side that is causing an issue for that specific location.
A more costly way to deal with this would be to set up network and workstation monitoring software and look for hiccups. But this is not very affordable for most, and in the past I have pretty much done pretty well diagnosing network problems by pinging from strategic locations and sending the information from the pings to a log file with date/time stamps to look at further. One affordable tool I did buy was Alert Ping Pro
http://www.bestshareware.net/alert-ping.htm which allowed for me to be notified right when problems are detected by having it trigger a program I wrote and compiled as exe which would call me on my cell phone and play an audio track looped for pointing out where problem area is.
One last thing to check would be the event logs of the systems that have had this issue to see if there is anything there that points out any issues. as well as I had an accounting system once that had random issues that I thought was a bug with an accounting program, but it ended up being a 1GB RAM stick that was failing in the 628MB range according to memtest86, and all other programs were not memory hogs and so the system ran fine, but when the memory hog accounting program ran a bug like problem would pop up in regards to a function call. Swapping out the RAM stick after finding it was problematic fixed the issue with that HP SFF Business class workstation.
