Rock Quality Designation (RQD)

In 1964 Don Deere proposed a way of putting a number on how broken a rock mass is, using something every drilling programme already produced: the core box. Lay out a run, measure every intact piece that is at least 100 mm long, add those lengths, divide by the length of the run and call it a percentage. That is the whole of it, and sixty years later it is still the first number anyone asks for.

It survives because it is cheap, quick, and repeatable between two people who have both been shown how to do it — not because it is a good measure of anything. RQD ignores joint orientation entirely. It counts how broken the rock is and has no way of knowing which way the pieces would slide. The same 70% describes a mass whose single joint set dips harmlessly into the wall you are about to cut and a mass whose joint set daylights straight out of it, and one of those is a wedge waiting to go. In slope work, where orientation is usually the entire problem, an RQD on its own is close to useless.

It also saturates. Every piece longer than 100 mm counts the same as every other, so a rock jointed at 110 mm spacing and a rock jointed at four metres both score 100%. Above that ceiling the index has stopped measuring. And the 100 mm threshold is arbitrary and behaves as a cliff rather than a slope: a run of pieces averaging 95 mm scores zero, and the same rock recovered a few millimetres kinder scores a hundred. Nothing about the ground changed.

Three logging errors account for most of the bad RQD in the world, and it is worth naming them because two of the three flatter the rock. Measuring along the wrong axis is the classic: piece length is measured along the CORE AXIS, down the centreline, not along the longest edge of a piece that has been cut off at an angle. A joint dipping 30° to the core produces a wedge whose long edge is far longer than its centreline, and measuring that edge inflates every piece in the run. Counting drilling breaks as natural discontinuities is the opposite error and the only conservative one in the subject: fresh, rough, matching fractures made by the bit or by rough handling are not joints, and the pieces they separate should be fitted back together and logged as one. Using the recovered length instead of the drilled length is the third, and it is the worst, because core lost in the hole is core that scored zero — and it is nearly always the most broken and most important rock in the run.

What RQD is genuinely good for is being an input. It is the first term of Barton's Q, it is one of the six parameters of Bieniawski's RMR, and in both of those it is surrounded by other terms that supply the things it cannot see. Read on its own it is a rough sorting of ground into Deere's five bands — excellent above 90, good 75 to 90, fair 50 to 75, poor 25 to 50, very poor below 25. Read as the first line of a classification it is doing the job it was designed for.