Lonsdale addresses the generalities in Tree Hazard Assessment and Management s. 5.2.1.2 but in the end he says only that "A visual inspection by an experienced person can help to decide whether the area of bark-to-bark contact is geat enogh to pose a hazard". When doing QTRA assessments I have to put a range of probabilities of failure on compression forks regularly. Hard as this may seem, if you think in orders of magnitude of probability it is good enough, but it has taken years of call-outs to storm breakages and surveying thousands of trees that have obviously stood up to those same storms to get an idea of what is an acceptable fork and what is not.
Mattheck and Breloer in The Body Language of Trees, your best chance of an overall biomechanical and numeric approach to assessing strength, do not even try to quantify strength. There are intersting comments about 'cupboard door failures' which are a reminder that the alignment of the bark inclusion in a fork relative to wind loads can be significant. Thus it is perhaps more likely that in southwesterly winds a north-south inclusion is more likely to fail than a northwedst-southeast inclusion. But only if the tree has not adapted to endemic south westerlies.
That book also suggests that forks with leaning codominant stems are more likely to be weak by the jacking open of compression forks due to annual wood increments in the fork itself.
I was up a beech today, c. 25m high with a deep pocket included fork near the base. While thinning the smaller (400mm dia at the base) crown at a height of about 15 metres, and anchored on it and the main substem, I witnessed to my great alarm the substem moving perhaps 300mm awy from and towards the main stem only under my body weight (c. 70kg). Yet it was protected from wind by an adjacent tree and the fork was almost perfectly sw-ne. The tree had to be 150 years old and had probably had this inclusion for 130 of those.
In short, there's no rule. But I have a feeling that in the same position with similar dimensions a Norway Maple or a Cedar would have failed 50 years ago. On the same site there were 2 other beech I had climbed past huge compression fork tear wounds in a similar orientation, one was very exposed, the other had been exposed at maturity to fresh loads by the removal of adjacent trees. Yet no 'cupboard door' effect as the substems must have been enormous with so much dead weight that sudden failure led to instant vertical drop of the limbs with oddly minimal tearing beneath. Both trees were thinned by me, and I had little concern for the wound position being weak. n the contrary the reaction wood around the wounds was fabulous, if a little ramshorned.