Indiacon CivilSubscribe

Textbook › Reinforced Concrete Design › Chapter 1 › Lesson 1.3

1.3 Characteristic loads and strengths

Reinforced Concrete Design Chapter 1: Design philosophies Under review — not yet three-checked Author: Indiacon

Learning objectives

After this lesson you should be able to:

  • Define characteristic strength and characteristic load as statistical fractiles and explain why design starts from them rather than from average values.
  • Estimate the characteristic strength of concrete from test results using fck = fm − 1.645 s, and check whether test results of concrete and steel meet a specified grade.
  • Find characteristic dead and imposed loads for a floor and convert them into service and design loads.

Theory

No two cubes from the same concrete give the same strength, and no two floors carry exactly the same load. Design therefore cannot use 'the' strength or 'the' load; it uses values chosen so that only a small, stated proportion of cases is worse. These are characteristic values. IS 456:2000 cl. 36.1 (and cl. 6.1.1 for concrete) defines the characteristic strength as the value below which not more than 5 per cent of test results are expected to fall, and cl. 36.2 defines the characteristic load as the value with a 95 per cent probability of not being exceeded during the life of the structure.

Concrete strengths are close to normally distributed. In a normal distribution 5 per cent of values lie more than 1.645 standard deviations below the mean, so fck = fm − 1.645 s, where fm is the mean and s the standard deviation of the results. Read the other way, a supplier who wants to meet a grade must aim at a target mean strength above fck by 1.645 s (rounded to 1.65 s in Indian mix design), so better control (smaller s) allows a lower mean and a cheaper mix. The grade name carries fck directly: M25 means a characteristic 28-day cube strength of 25 N/mm².

Characteristic strength: the 5 per cent fractilecube strength (N/mm²)frequency of resultsfm (mean)fck1.645 s5 % of resultsIndiacon original, not to scale
Figure 1.3: Frequency curve of cube strength: the characteristic strength fck lies 1.645 standard deviations below the mean fm, so that 5 per cent of results (shaded tail) fall below it. Indiacon original, not to scale.

For reinforcing steel IS 456:2000 cl. 36.1 and cl. 5.6.3 take the characteristic strength as the minimum yield stress or 0.2 per cent proof stress specified in the steel standard, IS 1786 for deformed bars: the grade number of Fe 415, Fe 500 or Fe 550 is that minimum in N/mm². Steel is made under factory control and its scatter is much smaller than that of concrete, which is one reason γm for steel is smaller.

For loads, real statistics are rarely available, so IS 456:2000 cl. 36.2 adopts the values of the loading codes as characteristic: dead loads from IS 875 (Part 1), imposed loads from IS 875 (Part 2), wind from IS 875 (Part 3), snow from IS 875 (Part 4) and seismic forces from IS 1893. Dead loads come from dimensions and unit weights (reinforced concrete 25 kN/m³, IS 456:2000 cl. 19.2.1) and vary little; imposed loads vary a great deal with use, which is why the load factors of lesson 1.2 treat them more cautiously in many codes.

Derivation

  1. Let strength f be normal with mean fm and standard deviation s. The standardised variable z = (f − fm)/s is standard normal.
  2. The characteristic strength is the 5 per cent fractile: P(f < fck) = 0.05, so Φ((fck − fm)/s) = 0.05.
  3. From standard normal tables Φ(−1.645) = 0.05, so (fck − fm)/s = −1.645 and fck = fm − 1.645 s.
  4. The target mean for a mix follows by rearranging: fm,target = fck + 1.645 s; with s from about 30 or more results the estimate is reliable, otherwise an assumed s is used.
  5. For a load Q with mean μQ and standard deviation σQ, the characteristic load is the 95 per cent fractile, Qk = μQ + 1.645 σQ, the mirror image of the strength rule.
  6. Probability of a low result: for any value f0, P(f < f0) = Φ((f0 − fm)/s); for example a mix with fm = 33 and s = 4 N/mm² has P(f < 25) = Φ(−2.0) = 2.3 per cent.
Formulas, symbols and SI units
  • Characteristic strength of concrete: fck = fm − 1.645 s (N/mm²)
  • Target mean strength: fm = fck + 1.645 s (N/mm²), rounded to 1.65 s in Indian mix design practice
  • Sample mean and standard deviation: fm = Σf/n; s = √[Σ(f − fm)²/(n − 1)] (N/mm²)
  • Characteristic load: Qk = μQ + 1.645 σQ (kN/m² or kN)
  • Characteristic strength of steel: minimum yield or 0.2 per cent proof stress of the grade, fy (N/mm²)
  • Dead load of a slab: g = γ t (kN/m²), with γ = 25 kN/m³ for reinforced concrete and t in m
Assumptions and limits
  • Test results come from one population: the same mix, materials, curing and testing, sampled at random.
  • The normal distribution describes the scatter well near the 5 per cent tail; with few results the estimates of fm and s, and so of fck, are uncertain.
  • Characteristic loads in Indian practice are the code values of IS 875 and IS 1893, not statistically derived fractiles, because load statistics are not generally available.
  • The steel's characteristic strength is the specified minimum of its grade, not the mean of mill test results.
  • Limits: the 1.645 rule does not replace the code acceptance criteria for concrete on site, which use their own sample sizes and margins; characteristic values do not cover gross errors such as wrong batching or missing bars.

Codes and standards

Indian Standards are cited by number and clause only; no clause text, tables or figures are reproduced. Other standards are listed for comparison, not ranked.

  • IS 456:2000 cl. 6.1.1 and cl. 36.1 (characteristic strength: not more than 5 per cent of results below it), cl. 6.1.2 with Table 5 (minimum grade; M 20 for reinforced concrete in mild exposure), cl. 5.6.3 (Es = 200 kN/mm²; characteristic yield strength of steel is the specified minimum yield or 0.2 per cent proof stress), cl. 36.2 (characteristic load: 95 per cent probability of not being exceeded; loading codes adopted as characteristic), cl. 19.2.1 (unit weights: 24 kN/m³ plain and 25 kN/m³ reinforced concrete).
  • IS 875 (Part 1) for dead loads, IS 875 (Part 2) for imposed loads (2.0 kN/m² for residential rooms), IS 875 (Part 3) for wind and IS 875 (Part 4) for snow; IS 1893 for earthquake forces; IS 1786 for high strength deformed steel bars (grades such as Fe 415, Fe 500 and Fe 550, and the D grades with higher ductility).
  • Indian, European and American practice are listed side by side and not ranked: the European system (EN 1990, EN 1991 for actions, EN 206 and EN 1992-1-1 for concrete) defines characteristic strength as the 5 per cent fractile and names classes by cylinder and cube strength; the American system (ACI 318) specifies a cylinder strength and sets required average strengths from the standard deviation of the producer's results. Their numerical rules are not reproduced here.

Worked examples

Example 1 (exam style: characteristic strength from cubes)

Problem. Ten 28-day cube results (N/mm²) from an M25 pour are 31.2, 28.4, 35.6, 30.1, 33.8, 27.5, 32.9, 29.7, 34.4 and 31.0. Estimate the mean, the standard deviation and the characteristic strength. Separately, if a supplier's mix has fm = 33 N/mm² and s = 4 N/mm², what proportion of results would fall below 25 N/mm² and below 20 N/mm²?

  1. Sum = 314.6, so fm = 31.46 N/mm².
  2. s = √[Σ(f − fm)²/9] = 2.66 N/mm².
  3. Descriptive estimate only: fm − 1.645 s = 31.46 − 1.645 × 2.66 = 27.09 N/mm². This describes the ten results; it is not an acceptance decision. Formal acceptance follows IS 456:2000 cl. 16.1 and Table 11: the mean of each group of four non-overlapping consecutive results (here 31.33 and 30.98 N/mm² for the first two groups) and each individual result are checked against the limits in the current amended code.
  4. For the supplier's mix: z = (25 − 33)/4 = −2.0 and Φ(−2.0) = 0.023, so about 2.3 per cent of results fall below 25 N/mm².
  5. Below 20 N/mm²: z = −3.25 and Φ(−3.25) = 0.00058, about 6 in 10 000.

Answer. fm = 31.5 N/mm², s = 2.66 N/mm², fck ≈ 27.1 N/mm² (meets M25); 2.3 per cent below 25 and 0.06 per cent below 20 N/mm²

Example 2 (site style: steel test results)

Problem. Eight tensile tests on 16 mm bars sold as Fe 500 give 0.2 per cent proof stresses of 548, 562, 531, 575, 544, 556, 539 and 567 N/mm². Assume the purchase condition is that every result must be at least the grade value. Check the bars, estimate the 5 per cent fractile of their proof stress, and state which value the designer uses.

  1. Every result is at least 531 N/mm², above the specified minimum of 500 N/mm², so the batch meets the grade on proof stress (other properties such as elongation are checked separately).
  2. Mean = 552.75 N/mm² and s = 14.91 N/mm².
  3. 5 per cent fractile = 552.75 − 1.645 × 14.91 = 528.2 N/mm², comfortably above 500.
  4. The designer still uses fy = 500 N/mm² and the design strength 500/1.15 = 434.8 N/mm², never the test mean.

Answer. All results ≥ 500 N/mm²; mean 552.8, s 14.9, 5 per cent fractile 528 N/mm²; design uses fy = 500 N/mm² (0.87 fy = 435 N/mm²)

Example 3 (characteristic, service and design loads on a floor)

Problem. A residential floor slab is 150 mm thick with 1.0 kN/m² of finishes and spans 4.0 m one way, simply supported. Find the characteristic dead and imposed loads, the service and design loads and moments per metre width. Also, if a survey found the peak imposed load in such rooms to have a mean of 1.2 kN/m² and a standard deviation of 0.4 kN/m², what would its 95 per cent value be?

  1. Dead load g = 25 × 0.150 + 1.0 = 4.75 kN/m² (IS 456 unit weight 25 kN/m³); imposed load q = 2.0 kN/m² (IS 875 (Part 2), residential rooms).
  2. Service load = 4.75 + 2.0 = 6.75 kN/m²; service moment = 6.75 × 4.0²/8 = 13.5 kN m per metre.
  3. Design load = 1.5 × 6.75 = 10.125 kN/m²; design moment = 10.125 × 4.0²/8 = 20.25 kN m per metre.
  4. Survey-based 95 per cent value = 1.2 + 1.645 × 0.4 = 1.86 kN/m², just below the code value of 2.0 kN/m², which is the one used in design.

Answer. g = 4.75 and q = 2.0 kN/m²; service 6.75 kN/m² (13.5 kN m/m); design 10.125 kN/m² (20.25 kN m/m); survey 95 per cent value 1.86 kN/m²

Step-by-step procedure

  1. Identify each material grade; for concrete fck is the number in the grade name, for steel fy is the grade number.
  2. For test data, compute the mean and the sample standard deviation, and estimate the 5 per cent fractile with 1.645 s.
  3. Compare with the specified grade and, for acceptance, apply the code acceptance criteria, not the fractile alone.
  4. Compute dead loads from member sizes and unit weights and take imposed, wind, snow and seismic loads from the loading codes as characteristic values.
  5. Form service loads (characteristic values with serviceability factors) and design loads (characteristic values with γf) for the limit states of lesson 1.2.
  6. Record the characteristic values used on the drawings or design basis so that site testing can be checked against them.

Site notes

  • Cube results are only meaningful if the cubes are made, cured and tested properly; a low result from a badly handled cube says more about the handling than the concrete.
  • Keep a running record of cube results for each mix; the standard deviation from 30 or more results tells the supplier how much margin the target mean really needs.
  • Ask for mill test certificates for every steel lot and check them against the grade, including elongation for D grades; spot-check with independent tests on important work.
  • Check that floor finishes, screeds and partitions on site match the dead loads assumed in design; extra screed thickness is a common hidden overload.
  • Change of use (for example a residence turned into a store or office) changes the characteristic imposed load; it needs a design check before the new use starts.
Common mistakes
  • Using the average cube strength as fck, which overstates the strength by about 1.645 s.
  • Using the population formula (dividing by n) for s with a small number of results; use n − 1.
  • Designing with the mean steel test strength instead of the grade value.
  • Forgetting finishes, partitions or services in the dead load.
  • Treating 1.65 s as a safety factor: it locates the 5 per cent fractile, and the partial factors are applied on top of it.
Quick check

For good site control s is about 3.5 to 5 N/mm², so the mean cube strength is typically 6 to 8 N/mm² above the grade (for example about 31 to 33 N/mm² for M25). If the mean of a large set of results is less than the grade plus about 1.645 × 3.5 = 5.8 N/mm², more than 5 per cent of results are likely to fall below the grade and the mix or the control needs attention.

Summary

  • Characteristic strength is the 5 per cent fractile: not more than 5 per cent of results fall below it.
  • For normal scatter fck = fm − 1.645 s, and a mix must target fm = fck + 1.645 s.
  • Characteristic steel strength is the specified minimum yield or 0.2 per cent proof stress of the grade.
  • Characteristic loads are 95 per cent values; in Indian practice they are the values in IS 875 and IS 1893.
  • Design values are formed from characteristic values with partial safety factors (lesson 1.2).

Practice questions

Tap a question to see the answer.

1A set of cube results has mean 35 N/mm² and standard deviation 4 N/mm². Estimate fck.

35 − 1.645 × 4 = 28.4 N/mm².

2What target mean strength is needed for M25 if s = 4 N/mm² (using 1.65 s)?

25 + 1.65 × 4 = 31.6 N/mm².

3If a mix is produced exactly at fm = fck + 1.645 s, what proportion of results falls below fck?

5 per cent.

4What is the self-weight of a 200 mm reinforced concrete slab?

25 × 0.200 = 5.0 kN/m².

5What does the grade name Fe 500 tell the designer?

The characteristic strength: a specified minimum yield or 0.2 per cent proof stress of 500 N/mm².

6Why are IS 875 values used as characteristic loads instead of measured 95 per cent values?

Load statistics are not generally available, so IS 456:2000 cl. 36.2 adopts the loading code values as the characteristic loads.

References

Standard texts, listed in no order of merit. Our explanations, examples and diagrams are original; nothing is copied from these books.

  • S. U. Pillai, D. Menon, Reinforced Concrete Design, Tata McGraw Hill (now McGraw Hill Education (India)), 3rd edition, 2009.
  • P. C. Varghese, Limit State Design of Reinforced Concrete, Prentice Hall of India (PHI Learning), 2nd edition, 2008.
  • J. K. Wight, J. G. MacGregor, Reinforced Concrete: Mechanics and Design, Pearson, 6th edition, 2012.
  • R. Park, T. Paulay, Reinforced Concrete Structures, John Wiley and Sons, 1975.
Suggest a correction

If something on this page is wrong, tell the editor. Your message is sent privately and is not published.

Sent privately to the Indiacon editor; not published.