Clustering, Can I Force A Cluster?

I've been using some of the data mining capability in SQL 2005, and have been very impressed. But I've mostly been using decision trees/neural nets, and now have a problem that would seem to require clustering, but I'm not sure how to approach it...

Have list of parts by serial number and part number, and in a data cube have crunched a bunch of data to give me each parts' relative reliability.

So what I can do so far is to ask the cube "for a certain part number, which specific serial numbers are our 'bad actors'", ie have an unacceptably high failure rate compared to the average, by standard deviation.

The next logical question would be... what factors do these particular "bad" parts have in common? Environmental exposure, usage rates, a particular repair facility, etc etc. I have that type of raw data in other tables, no problem.

So what it seems like I would want to do is force a clustering algorithm to choose a certain segment of parts as a cluster and then use DM to determine what characteristics those parts share. Then we could see if there was any corrective action we could take, or whether we should just chuck those parts in the bin (if the DM didn't find any reasonable relationships between the parts). But I don't know enough about the clustering/association type DM to figure it out.

How does the group think I could approach this?

Geof