242-MDC : Multi-decade creep and shrinkage of concrete: material model and structural analysis

Technical Committee 242-MDC

General Information

Chair: Prof. Zdenek P. BAZANT
Deputy Chair: Prof. Milan JIRASEK
Activity starting in: 2010
Activity ending in: 2015
Cluster C

Subject matter

  Material: Concrete
  Structures: Concrete structures, durability and service life included.
  Level: Theoretical analysis with evaluation of published experimental results and observations on structures.
  Novelty: Not a new subject but a new approach to the subject, based on an improved theory and recent observations on structures.
  Scope: Extended to multi-decade.

Phenomenon and Problem that Needs to Be Addressed:

The prestressed box girder in Palau, of world-record span 241 m, deflected within 18 years by 1.61 m, instead of 0.3 m (versus camber) expected in design, and the prestress loss reached 50%. These facts would probably have never become known if ill-conceived remedial prestressing did not lead to delayed collapse with fatalities. 12 years after the collapse, a resolution on engineering ethics by the Structural Engineers World Congress led to the release of technical data, first only to Northwestern Univesity. Their sophisticated analysis showed that the use of a better prediction model for creep and shrinkage, along with better (three-dimensional) finite elements analysis and moisture diffusion analysis, could have predicted the observed values of deflections and prestress loss correctly. These finding, first reported at the last CONCREEP-8 conference in Ise-Shima, Japan, have raised alarm.
Subsequent efforts of the proposer and several prospective committee members revealed that similar excessive deflections (though without collapse and without fatalities) have been observed on numerous bridges, especially in Europe and Asia (in the US, there are hardly any examples since the material for most large-span bridges was steel until very recently). Although companies are generally loath to publicizing such negative information on their structures, it seems that information could be obtained by personal contacts for a number of such bridges in Japan, Korea, and perhaps also Europe.
These recent revelations showed that the creep and shrinkage predictions models used by all engineering societies, as well as the methods of analysis used in industry (based on one-dimensional beam-type finite element programs, and on neglect of moisture diffusion phenomena) are hopelessly inadequate for multi-decade predictions. The largest world-wide database (the 2009 ITI database), recently compiled at Northwestern University by enlarging the previous RILEM database (compiled in a previous RILEM TC chaired by the proposer) contains almost 20,000 data points, but only about 3% are for load durations exceeding 8 years, and only about 0.5% for load durations exceeding 13 years.
Thus there is no way to verify and calibrate a multi-decade creep and shrinkage prediction model from the world-wide database alone. It is inevitable to rely also on:
  a) mathematical-physical basis of the phenomenon, which recently came to be understood far better (including its nano-poro-mechanics aspects) and, especially,
  b) interpretation or inverse analysis of excessive deflections and excessive prestress losses of large creep-sensitive structures.

Prediction models for long-time (especially multi-decade) creep and shrinkage of concrete. Multi-decade deflections, prestress loss and damage in creep-sensitive concrete structures, which particularly include large-span bridges (especially prestressed box girders), very tall buildings, nuclear containments, large span shells and large-span roof beams).

Terms of reference

Time needed:
No more than 5 years is allowed by the rules of RILEM.

Because of the recent developments in the underlying theory, in the methods of unbiased statistical data evaluation, and in the rational interpretation of long-time observations on structures, it will be crucial for success to have as members the top specialists in mathematical modeling of creep and in statistical evaluation of data. It is expected that much of the work will be done by electronic correspondence. Conference calls will be held as well. The committee meetings will be planned, as much as possible, in conjunction with the conferences with maximum participation by committee members, to minimize the travel costs and raise attendance, and the semi-annual RILEM conferences will be preferred for the committee meetings.

Committee Work:
Included: Bibliographical research? -Yes.
Computer processing? - Yes.
Special Equipment? - No, unless computers count.
Reference materials? - Yes.
Round robin tests? - Not yet clear but probably not.

Detailed working programme

1) Comprehensive and detailed literature search.
2) Personal contacts with some selected firms and organizations soliciting data on excessive deflections and damage of structures under their competence (expected are great difficulties, since companies and owners generally dislike revealing data on problems with their structures, and legal agreements on the sealing of technical data after a court litigation usually stand in the way).
3) Selection of computer method of structural creep analysis, including the effects of moisture diffusion, heat conduction, cracking, and aging due to hydration.
4) Computer simulations, based mainly on finite element programs, of multi-decade response of selected structures on which good data have been selected, using various prediction models for creep and shrinkage.
5) Adjustments of model parameters so as to obtain close fits of multi-decade deflections and prestress losses.
6) Possibly, inverse finite element analysis of structures, which is a much more powerful approach to the foregoing task. This will be highly demanding both computationally and theoretically and will depend on the success in obtaining externally sponsored research funding.
7) Selection of optimum creep and shrinkage prediction model to be recommended.
8) Study of the best way to quantify prediction uncertainty and extrapolation of short-time tests.
9) Recommendation of a method of predictive multi-decade analysis of creep-sensitive structure (RILEM Recommendation).
10) Publicize the results in ACI, fib, IABSE, at CONCREEP conference, and perhaps also in JSCE, KCI, TCI, etc

Technical environment

Dialog among various leading research institutions, design firms, forensic engineering firms (such as WJE, Inc., and Exponent, Inc.); contacts with other RILEM committees on other subjects, if pertinent; and contacts with ACI (Committee 209 on Creep and Shrinkage, of which the proposer has been a long-time member), with IA-CONCREEP, (co-founded by the proposer) and with ‘fib’.

Creep and shrinkage has been one principal interest in RILEM for a long time. The proposer has previously chaired two RILEM committees on this subject. Recent advances in computer modeling, the underlying theory, as well as recent revelations of excessive deflections and prestress losses call for revisiting this subject, to benefit the structural engineering profession.

Expected achievements

I. RILEM Recommendation on:
a) a realistic prediction model for multi-decade creep and shrinkage;
b) an effective method of multi-decade extrapolation of short-time tests.
c) a realistic method of analysis of creep and shrinkage effects in large creep-sensitive structures.
d) method of predicting the multi-decade uncertainties, particularly the coefficient of variation and confidence limits.
e) possibly, development of a website presenting the database and some key computer programs.
Note: The depth and extent of each achievement will depend on the success with external funding.

II. A symposium on the subject at the triannial CONCREEP conference (co-sponsored by RILEM). The next conference, which will be organized by president Franz-Josef Ulm at MIT in 2011, could feature such a Symposium, but the final one will have to held at anniversary CONCREEP-10 in 2014 (perhaps in Prague).

Group of users

a) Engineers in design and construction firms;
b) Academic and institutional researchers in the field;
c) Members of various engineering society committees preparing design recommendations or codes.

Specific use of the results

The best use of the results would be for modern prestressed box girders.
According the many State Departments of Transportation in the US, bridges should last at least 100 years. Even more is expected from very tall buildings. It is now clear that current design methods do not ensure such a lifetime. Although no precise information exists in the archival journals, and much of what exists is hearsay, hundreds of large prestressed box girder bridges deflect far more then expected and suffer cracking damage.
Typically, problems arise after 20 years. At that time nobody is held liable, especially since the investigating agencies usually discover several other simultaneous deficiencies (e.g., “poor quality” of construction or maintenance) on which the problem can be blamed entirely, instead of blaming it on the design recommendations of engineering societies.
Needless to elaborate, shortening of design life from 100+ years to about 20 represents an enormous economic loss, amounting probably to billions of dollars.
Occasionally, as in the case of the bridge in Palau, incorrect remedies may lead to injuries and loss of life. In rare instances (shells, columns), long-time creep can directly cause collapse, due to creep buckling.