10 Common Mistakes When Calculating Rock Mass Rating (RMR)
The Bieniawski 1989 Rock Mass Rating system looks deceptively simple. Six parameters, six lookup tables, an addition step, and an orientation adjustment. Yet in practice, even experienced geotechnical engineers make systematic errors that can shift the final RMR value by 10 to 30 points — enough to push a rock mass from one class to another and produce dramatically different support recommendations. This guide collects the ten most frequent mistakes encountered in field reports, design calculations, and student assignments. Each mistake is explained, the correct procedure is described, and the consequences of getting it wrong are quantified. If you want to follow along with a working example, our RMR calculator implements all six parameters with the correct lookup logic.
Mistake 1: Confusing UCS and Point Load Index
Parameter 1 of the RMR system rates the uniaxial compressive strength (UCS) of intact rock material. The Bieniawski 1989 table provides two parallel rows: one for UCS measured directly through laboratory testing, and one for the point load index Is(50) measured through the simpler point load test. The mistake is to take an Is(50) value, multiply it by an arbitrary conversion factor (typically 20 to 24), and then use the UCS row of the table.
This double conversion introduces avoidable error because the conversion factor between Is(50) and UCS is not constant — it varies from about 12 for weak shales to nearly 30 for very strong granites. The Bieniawski table was calibrated against direct measurements, so using the Is(50) row with the actual measured Is(50) value gives the correct rating. If you only have Is(50) data, use the Is(50) column directly. If you have UCS data, use the UCS column. Never convert one to the other unless absolutely necessary, and never apply the converted value to the wrong column. For a complete walkthrough of all six parameters, see our guide on RMR parameters explained.
Mistake 2: Calculating RQD Incorrectly from Core
Rock Quality Designation (RQD) is defined by Deere as the percentage of intact rock pieces longer than 100 mm in a core run, divided by the total length of the core run, expressed as a percentage. The most common error is to include short or broken pieces in the numerator. Only sound, intact pieces measured along the centerline of the core, with each piece at least 100 mm long, count toward the numerator. Mechanically broken pieces caused by drilling stress, handling damage, or core barrel jamming should be reconstructed and measured as if continuous.
A second common error is to measure the longest dimension of irregular pieces rather than the centerline length. The ISRM standard requires centerline measurement to ensure consistency between operators and projects. A third error is to count pieces from a core run that includes drilling refusal or core loss without adjusting the denominator appropriately. The denominator should be the actual length of the core run, including any zones of total core loss, which themselves represent very poor rock quality. Underestimating the denominator by ignoring core loss zones inflates the RQD and produces an optimistic rating. For surface or wall mapping where core is unavailable, RQD can be estimated from the volumetric joint count using Palmstrom's formula RQD = 115 − 3.3 Jv, but this estimate carries higher uncertainty than direct core measurement.
Mistake 3: Misinterpreting Joint Spacing
Parameter 3 rates the spacing of discontinuities. The mistake here is to use the average spacing of all joint sets, calculated as a simple arithmetic mean. This approach can mask the influence of a single closely spaced set that dominates rock mass behavior. The Bieniawski 1989 system requires the engineer to identify the dominant joint set — the one most likely to control failure modes given the excavation geometry — and use its spacing to enter the rating table.
For practical application, this usually means selecting the most closely spaced set, because the closely spaced set produces the smallest blocks and the highest fracture density. In some cases, the most relevant set may be the one whose orientation is most unfavorable relative to the excavation, even if it is not the most closely spaced. When in doubt, calculate the rating for each set independently and use the lowest rating, which corresponds to the worst case. Avoid averaging across sets unless the sets have very similar orientations and spacings, in which case averaging is approximately correct.
Mistake 4: Applying the Wrong Orientation Adjustment
Parameter 6 is the orientation adjustment, which subtracts up to 60 points from the basic RMR depending on whether the excavation is a tunnel (up to -12), a foundation (up to -25), or a slope (up to -60). The most common mistake — and the most dangerous — is to use the tunnel adjustment table when assessing a slope, or vice versa. This error can shift the RMR by 30 to 50 points, completely invalidating the resulting class and support recommendations.
The three application types use different tables because the failure mechanisms are different. Tunnel failures typically involve gravity-driven fall of wedges or roof loosening, where the relevant geometric parameter is the strike of joints relative to the tunnel axis and the dip of joints relative to vertical. Slope failures involve sliding along joints daylighting in the slope face, where the relevant geometric parameter is the angle between joint dip direction and slope face dip direction. Foundation conditions involve bearing capacity considerations along potential shear surfaces, where the relevant geometric parameter is the orientation of weak planes relative to the loading direction.
Always identify the application type before selecting the orientation adjustment, and use the corresponding Bieniawski 1989 table. If your project involves multiple types — for example, a portal where a tunnel intersects a rock slope — calculate RMR separately for each application using the appropriate adjustment. Our step-by-step RMR calculation guide walks through the orientation adjustment in detail.
Mistake 5: Mishandling the Joint Condition Sub-Parameters
Parameter 4 rates the condition of discontinuities through five sub-parameters: persistence, aperture, roughness, infilling, and weathering. Each sub-parameter has its own rating, and the five ratings are added to produce the total joint condition score (0 to 30 points). The mistake is to forget that the 1989 version uses sub-parameters at all, and instead apply the simpler descriptive scale from the 1979 version, which gives a single overall rating from 0 to 30 based on a qualitative description.
The 1989 sub-parameter approach is more rigorous because it forces the engineer to consider each aspect of joint condition independently. A joint can have low persistence (high rating) but be heavily weathered (low rating), and the additive scoring captures this nuance. Using the 1979 single-scale approach in a 1989 calculation can produce errors of 5 to 10 points on Parameter 4 alone. Always verify which version of the table you are using and apply it consistently. Some textbooks present both versions side by side, which is convenient for reference but a source of confusion if the engineer mixes them up.
Mistake 6: Mis-Rating Groundwater Conditions
Parameter 5 rates groundwater conditions through three indicators: inflow per 10 metres of tunnel length, joint water pressure relative to major principal stress, and a qualitative description (completely dry, damp, wet, dripping, or flowing). The mistake is to use only one of these three indicators and ignore the others, especially when they disagree. The Bieniawski 1989 table is structured so that the three indicators usually point to the same rating, but in practice they sometimes give different answers — for example, a tunnel with low inflow but high joint water pressure due to a confined aquifer.
When the three indicators disagree, use the worst case. A high water pressure with low visible inflow is more dangerous than dry conditions because it indicates pressurized water behind the face that may suddenly release during excavation. Similarly, a wet face with no measurable inflow may indicate water trapped in a closed joint network ready to drain catastrophically when intersected. Conservative selection of the groundwater rating protects against these scenarios. For projects with active dewatering or grouting, document the assumed groundwater condition both with and without dewatering, since the rating may need to be revised mid-construction.
Mistake 7: Treating RMR as a Precise Number
RMR is reported as an integer between 0 and 100, which gives an illusion of precision that the underlying data does not justify. A rock mass might genuinely warrant any value from 38 to 45 depending on which observer makes the assessment, which face is measured, and how the dominant joint set is identified. Treating the result as a single precise number — for example, "RMR = 42, therefore Class III" — ignores this real uncertainty.
Best practice is to report a band of values and the confidence level. A more honest reporting format is: "RMR = 38 to 45 (mean 42), Class III with portions of Class IV." Design decisions should then be based on the lower end of the band, not the mean, particularly for safety-critical structures. This approach also makes future comparisons across logged faces more meaningful, because variations within the natural uncertainty band are not mistaken for actual changes in rock quality. Engineering judgment is preserved as part of the workflow rather than hidden behind a single false-precision number.
Mistake 8: Ignoring Direction of Drive in Anisotropic Rock
The orientation adjustment in Parameter 6 depends on the geometric relationship between joint orientations and the excavation direction. In bedded rock or rock with a single dominant joint set, the rating can vary by 12 to 25 points depending on whether the tunnel is driven with the strike or against it, and whether joints dip toward or away from the working face. The mistake is to assess RMR for a tunnel as if it were direction-independent, producing a single rating that is correct for one direction of drive but incorrect for the reverse direction.
Always document the direction of drive used in the assessment, and recalculate RMR if the direction changes during construction. For a tunnel with two faces advancing toward each other, calculate RMR independently for each face direction. For mining excavations where the orientation may change repeatedly, develop a series of RMR values corresponding to the principal advance directions encountered. Failing to do this can lead to systematic over-support in one direction and under-support in the other, with corresponding cost and safety implications.
Mistake 9: Applying RMR Outside Its Calibration Range
The Bieniawski 1989 RMR system was calibrated against case histories of civil tunnels with spans roughly between 5 and 20 metres in moderately to well-jointed rock. Applying the system to dramatically different conditions — very large caverns (spans greater than 25 metres), very small openings (less than 3 metres), evaporites, swelling clays, or weak weathered rock — produces results that may not reflect actual ground behavior. The mistake is to apply RMR mechanically to any rock mass without considering whether the conditions are within the system's intended range.
For applications outside the calibration range, RMR should be supplemented or replaced by methods better suited to the specific conditions. Large caverns may require numerical modelling and the Hoek-Brown failure criterion. Weak rock may require the Geological Strength Index (GSI) extension developed by Hoek. Mining excavations may require the Mining Rock Mass Rating (MRMR) developed by Laubscher, which includes adjustments for blasting damage, weathering during exposure, and stress changes due to excavation. Always check whether RMR is appropriate before reporting it as the primary classification, and document any caveats in the geotechnical report.
Mistake 10: Failing to Recalculate as New Data Becomes Available
RMR is not a one-time calculation. As exploration drilling proceeds, additional core becomes available, more exposures are mapped, and the geological model is refined. Each new piece of information can change the RMR rating for the same rock mass — sometimes substantially. The mistake is to compute RMR once during preliminary design and never revisit it, even when significant new information becomes available during detailed design or construction.
Build RMR recalculation into the project workflow. At each project phase — feasibility, preliminary design, detailed design, and construction monitoring — review the available data, recalculate RMR for the relevant zones, and document any changes. If construction face mapping reveals that the actual rock mass is consistently better or worse than the design RMR, update the support specifications and inform the design team. RMR is most valuable when it tracks actual conditions throughout the project, not when it is treated as a fixed input established during initial site investigation. For practical examples of how RMR recalculation looks in real projects, see our RMR worked examples page.
Putting It All Together
Avoiding these ten mistakes requires more than memorizing the rating tables. It requires understanding why each parameter is included in RMR, how it relates to physical rock mass behavior, and what the practical limitations of the system are. The Bieniawski 1989 RMR is a powerful tool when applied carefully, but it is also a tool that rewards attention to detail and punishes carelessness. Engineers who treat it as a black box that converts numbers in to a rock class out will produce misleading results, especially in marginal conditions where small errors push the rating across class boundaries.
The good news is that all ten mistakes are easy to correct once you know what to watch for. Use the right table for the application type. Measure RQD on the centerline of intact pieces longer than 100 mm. Identify the dominant joint set rather than averaging. Apply the 1989 sub-parameter approach for joint condition. Report a band of values rather than a single integer. Recalculate as new data arrives. These habits, combined with sound engineering judgment and an understanding of the system's calibration limits, will produce RMR classifications that genuinely guide design decisions rather than mislead them.
For a step-by-step walkthrough of correct RMR calculation, see our how to calculate RMR guide. For comparison with the Q-System and recommendations on when to use which, see our RMR vs Q-System comparison. To run a calculation directly with the correct table logic, use the free RMR calculator.
Frequently Asked Questions
The most common mistake is misapplying the orientation adjustment in Parameter 6. Engineers often forget that the adjustment value depends on whether the excavation is a tunnel, foundation, or slope, and that these three categories use different correction tables. Using the tunnel adjustment for a slope assessment can underestimate slope instability by 30 to 50 RMR points, leading to dangerously optimistic stability predictions.
Yes, RQD can be estimated from surface scanline mapping using Palmstrom's volumetric joint count formula RQD = 115 − 3.3 Jv, where Jv is the number of joints per cubic metre. This is useful in outcrops, tunnel walls, or slopes where coring is not available. However, the result is an estimate, and direct measurement from core remains the standard reference method when core is available. ISRM recommends using both methods when possible and reporting any discrepancies.
The Bieniawski rating table accepts either UCS or point load index Is(50), but the engineer must use the correct row in the lookup table. The point load index is approximately UCS divided by a conversion factor of 20 to 25, depending on the rock type, but this is a generalization. Whenever possible, perform direct UCS testing on representative samples. If only Is(50) is available, use the Is(50) row of Bieniawski's table directly rather than converting to UCS first, because the conversion introduces additional error.
RMR should always be treated as a band of values, not a precise number. The class boundaries (20, 40, 60, 80) are sharp on paper but reflect a continuum of rock conditions in reality. A reported RMR of 41 is not meaningfully different from 40, even though they fall in different classes. Best practice is to report a range, such as RMR = 38 to 45 (Class IV/III boundary), and to base design decisions on the worst credible value rather than the mean.