The most recent ANZCA acte pain scientific evidence book has this to say about intra-articular analgesia (page 203-4).

... "there is evidence of a small benefit only of intra-articular local anaesthesia for postoperative pain relief after anterior cruciate repair (Moiniche et al 1999 level1)"

In addition, re: intra-articular morphine: "In clinical practice, morphine injected as a single dose into the knee intra-articular space produced analgesia that lasted p to 24 hors, bt evidence for a peripheral rather than a systemic effect was not conclusive (Gupta et al, 2001 Level 1; Kalso et al, 2002 Level 1). Confounding variables that hinder analysis inclded the pre-existing degree of inflammation, type of surgery, the baseline pain severity and the overall relatively weak clinical effect (Gupta et al, 2001 level 1). When published trials were analysed taking these confonding factors into consideration, including the intensity of early postoperative pain, the data did not support an analgesic effect for intra-articlar morphine following arthroscopy compaed with placebo (Rosseland, 2005 Level 1)."

This is a reversal of conclsions from the earlier pblication which did show a benefit.

In addition there is a small amount of evidence of chondrolysis associated with bupivacaine in shoulder surgery.

Variants that may not make it work would therefore include some of the confounders that weren't controlled for in the earlier studies- that is, things like the degree of pre-existing inflammation and the baseline pain severity. It would be interesting to see which of the 'types of surgery' variants had the most effect.

I guess that's what you're asking, eh.

After typing all that out I'm still going to post this answer. Then have a cup of tea and a good lie down.