By Wendell Dubberke
The American Concrete Institute (ACI) recently published a 78 page report titled "Visible and Invisible Cracking of Concrete" written by Richard Burrows. In this report, Dick discusses shrinkage micro-cracks that can occur during early concrete hydration. He proposes the use of a ring test method to identify those cements and/or mix designs that are susceptible to shrinkage micro-cracking.
When evaluating deteriorated concrete, identifying the sequence of events is most important. I have never been comforable with the idea that the excessive generation of expansive silica gel, ettringite and/or ice (or hydraulic pressure) are always primary events causing the concrete to crack. The pore system in high quality concrete (matrix) can easily handle moderate amounts of silica gel, ettringite or freezing water. Shrinkage micro-cracks, occurring during early concrete hydration, could later serve as conduits for moisture to more easily enter the concrete. Since most concrete deterioration mechanisms require moisture to function excessively, shrinkage micro-cracks could enhance later pcc problems related to expansion.
When evaluating the Duggan test method, one set of samples was monitored for growth throughout the whole test, including the pretreatment stage. Some of the samples contracted excessively during the pretreatment stage. Later some of the samples expanded due to excessive ettringite in the matrix.
Researchers at Iowa State University have evaluated a few cements using the ring test method. A limited number of ring test results indicate there may be a correlation between ring test results and Duggan test results. If there is a correlation, the ring test method would be the test method of choice because it is faster, easier and less expensive to operate. The ring test method works best when only water and cement are used in the mix. If a mix, containing aggregates, needs to be evaluated, then the Duggan test method should be used. The Duggan test method may be needed to identify cements associated with the excessive generation of ettringite, but it is possible that the cement chemistry responsible for later, excessive ettringite generation is the same chemistry responsible for initial shrinkage micro-cracking, particularly in low water mixes.
The Duggan test method was proposed by Ross Duggan (Chateauguay, Ontario, Canada) in the mid 1980s. The test method is approved and used to qualify concrete used by the railroad systems in North America. Ross Duggan, now deceased due to heart problems in 1998, tried to get an ASTM designation for his test method, but met considerable resistance from ASTM committee representatives.
The test method was originally envisioned as a measure of potential alkali-silica susceptibility of concrete. Old highway concrete or new concrete, prepared in a laboratory, could theoretically be tested. Later evaluations of the test method indicated the concrete expansions were related to the generation of ettringite rather than silica gel. Perhaps personnel at ASTM were confused as to the assignment of the test method.
In General, the Duggan test method consists of the 20 day measurement of five 1 inch by 2 inch concrete cylinders, drilled and removed from larger concrete specimens. Before measurement, the cylinders are pretreated by subjecting them to three cycles of dry heat (172 degrees F) and soaking in distilled water (room temperature) over a ten day period. Over the next 20 days, the cylinders are measured (to 1/10,000 of an inch) for expansion. A special rack fitted to a digital micrometer was used at the Iowa DOT. Between measurements, the 5 cyclinders from each concrete sample are stored in their own container filled with distilled water. At the conclusion of the measurements (20 days), the high and low measurements are discarded and the middle three are averaged to obtain the final test result expressed in percent growth. Throwing out the high and low readings was a change from what Duggan recommended. If the polished end on a concrete cylinder contains a void or has a reactive particle (chert or shale) at the point of measurement, it can have a disproportionate effect on test results. The samples were read in a horizontal position. The overall duration of the test is thirty days. Evaluation of the test at the DOT indicated that the test could be completed in twenty days (10 day read instead of 20) and still maintain adequate accuracy.
Duggan suggested that any expansion over .05% should be cause for concern. The North American railway system uses a .075% limit. For concrete that is used for structures that will always be high-and-dry, a higher limit could be used. For concrete that will come into contact with water, a Duggan test result over .075% should initiate a change in mix design or a change in concrete constituents. In some cases, an increase of water, in the mix design, could solve the problem.
A few years after the Duggan test was proposed, it was evaluated by concrete researchers from Canada. They found that the primary cause of concrete expansion, using this test method, was related to the generation of excessive ettringite in the concrete matrix. About this same time, some concrete investigators at the Iowa DOT were becoming concerned about the significant amount of fiberous, amorphous, hydrous, calcium sulfoaluminate (called ettringite gel for lack of a better name) filling the air voids in the early deteriorating portland cement concrete pavements (PCCP) built during the last 10 years. An evaluation of the Duggan test method was initiated at the Iowa DOT in the early 1990s. The evaluation of the Duggan test method proved successful (using concrete fabricated in the laboratory), and it was used to evaluate cements and water reducers in laboratory mixed concrete.
Other Iowa PCCPs, built during the last 20 years, and still performing adequately, also contained this ettringite gel in the air voids, but to a lesser degree. Older, durable Iowa PCCP contains only an extremely small amount of ettringite that is difficult to locate. Quantifying and qualifying ettringite gel is difficult because x-ray diffraction (XRD) will not work. The scanning electron microscope can show the elements in in ettringite gel as well as silica gel and can give indirect information about mineralogy. Concrete investigators at the Minnesota DOT have developed other techniques for measuring ettringite in concrete.
Is it OK to have air voids partially full of ettringite gel if the pavement is able to perform adequately? What is an adequate service life for PCCP? In Iowa, the best concrete aggregates are expected to perform for 30-40 years, even when deicing salts are used on the PCCP. Should air voids also function as reservoirs for excessive ettringite gel? Do you go after the source of the problem or do you use remedial techniques?
At DOT meetings (relating to early PCCP deterioration) some cement manufacturing representatives indicated that thier type I cement formulations were now semi shrinkage compensating (called expansive in the old days). Type K (and Type S, now obsolete) is a shrinkage compensating cement. The generation of substantial amounts of ettringite (gel?), at the proper time during concrete hydration, is the mechanism that makes it work. If all of the ettringite is formed while the concrete is still plastic, then cracking of the matrix will not occur because the expansive ettringite, even if it is excessive, will make room for itself before the concrete becomes brittle.
A low water/cement (W/C) ratio can hinder the total development of ettringite gel while the concrete is still plastic. If water becomes available at a later time and if it can get to the ettringite gel building components, ettringite gel will form. If the void system in the concrete matrix is insufficient for the volume of ettringite gel, cracking can occur.
During the 1960s, a Portland Cement Association (PCA) concrete researcher fabricated laboratory concrete that was expansive (due to delayed ettringite) by using a very low W/C ratio in the mix.
It is possible that the Duggan test is identifying those cements that have a high potential for developing large amounts of ettringite due to- (1) excessive use of grinding aids and/or (2) high amounts of ettringite building components (sulfur and aluminum) in their cement formulation. To be durable (long-term), maybe these cements should not be used with a low W/C ratio PCCP mix.
Back in the 1930s, W/C ratios of 0.6 or more were commonly used and even recommended. Recent trends are towards low W/C ratios that relate to higher early strengths. There seems to an assumption ("everybody knows") that high early strength equals long-term durability in concrete. This assumption may be true for structures that are high-and-dry (maybe).
Many concrete investigators say heat, applied during the hydration of concrete, is a requirement for the development of excessive ettringite that can crack the matrix. They dismiss the Duggan test method because it uses heat during the pretreatment phase of the test. Therefore, they say, expansions should be expected and consequently mean nothing. They are unconcerned about other concrete tests that use heat in their operation. In our meetings, cement company representatives also indicated a concern about the difficulty of formulating a cement that would consistently pass the test. Iowa DOT Duggan test results showed that some cement sources consistently failed the test and other cement sources consistently passed the test. One of the sources that consistently failed was used in failing Iowa PCCP. The failing PCCP also contained a lignosulfonate water reducer in the mix.
If the application of heat, during cement hydration, is required for the generation of excessive, expansive ettringite, how then can type K and type S expansive cement work without the application of heat? There are many stories floating around the concrete construction industry concerning the use of type S cement in construction work. Type S cement was formulated with extra sulfur and aluminum. The generation of substantial amounts of ettingite was intentional. The use of type S cement was abandoned because of poor performance. Additional heat was not applied to hydrating type S cement concrete, and yet it failed due to an ettringite related problem. The Portland Cement Association (PCA) initially promoted type S cement during the 1960s. Locating projects where type S cement was used along with its related service life is difficult to obtain.
Many concrete researchers also say that heat destroys ettringite. In a technical sense, it may be true, but realistically, if all that is happening is dehydration of ettringite, is this really destruction? Most likely, if water returns, ettringite will form again. Work at the Iowa DOT and Iowa State University (ISU), using the scanning electron microscpe (SEM), showed that heat (both dry and steam) did not alter the physical appearance of the ettringite gel in concrete air voids. It might be thought that dehydration would at least cause the ettringite to shrink, but it did not. Using QXRD to quantify crystalline ettringite before and after heating should show a difference. QXRD can not quantify the amorphous ettringite found in PCCP air voids.
Now, back to Iowa DOTs experience with the Duggan test method. Initial test results on field PCCP samples was not satisfactory. Prior work by other researches had alerted us to the fact that chlorides from deicing salts could influence test results. Also, if ettringite related expansion had already taken place, in older field PCCP, faulty test results could be expected.
Duggan test results, from laboratory prepared concrete, were very promising. Over 80 laboratory mixed concrete samples were eventually evaluated with the Duggan test method at the Iowa DOT laboratory.
After gaining some confidence in the Duggan test method (test results and operation), a variety of cements, water reducers and water/cement ratios were used in fabricating concrete specimens.
Complete chemical and physical tests were run on the cements and their results were correlated to Duggan test results. A similar regimen was applied earlier to the evaluation of the alkali/silica P214 test method. The P214 test method is now ASTM 1260. Fine-aggregates (sands) were also varied with the P214 evaluation. Periclase (MgO), from the cement, as determined by quantitative x-ray diffraction (QXRD) had an excellent, direct correlation to P214 test expansions. The cement alkalis correlated poorly with P214 expansions. More about this can be found on the alkali/silica reaction (ACR) link.
Duggan test results correlated best with elemental potassium and sulfur (0.9 and 0.8 r). Even though the correlations are good, they may be indirect. The amount of arcanite (K2SO4) in cement may relate to the amount of grinding aid added to the clinkers when manufacturing cement. The aggressiveness rather than the amount of grinding aid may also be a factor. The amount of arcanite (K2SO4) in cement relates to the amount of syngenite that can form in cement silos. The generation of syngenite, in cement silos, relates to flowability problems. Grinding aids, introduced into the system when the cement clinkers are pulverized, are used to maintain flowability in cement silos. Water reducers, plasticizers and retarders are surfactants and dispersents like grinding aids. Lignosulfonates can be used as water reducers or grinding aids.
It is unknown if Duggan test results directly relate to the amount of grinding aid in the cement because there is no easy way to qualify or quantify the grinding aid in cement. From the Duggan testing done so far, it appears that Duggan test results correlate directly to the amount of lignosulfonate (and others) water reducer incorporated in concrete mixes. A surfactant/dispersent overload in a low water/cement ratio concrete mix may relate to reduced PCCP service life.
If all of the ettringite forms (even if it is excessive) while the concrete is still plastic, the concrete will suffer no ettringite related problems, as evidenced by properly placed concrete made with type K cement. If, on the other hand, the concrete mix contains too much surfactant/dispersent, and in addition, the water/cement ratio is low, how can all of the ettringite generating components, in the cement, be expected to come together during the plastic stage of hydration? If the ettringite is precluded from forming during the plastic stage of hydration, wouldn't the concrete be subject to the generation of shrinkage cracks as it hardens? Later, wouldn't the shrinkage cracks allow for the easy movement of external water to the interior of the concrete and furnish water for the completion of ettringite generation, but unfortunately at a time when the concrete has already hardened? The heat used in the pretreatment phase of the Duggan test method may be generating micro-cracks in the matrix of the concrete cylinders. These micro-cracks could allow water to reach the ettringite building components, similar to what could happen to PCCP over a longer period of time.
In 1998, at one of our DOT meetings, a ready-mix owner showed pictures of a PCCP that had to be removed, shortly after construction, because of excessive shrinkage cracking. Is it possible that the mix design precluded the ettringite from forming early during initial hydration of this faulty PCCP.
As stated on other pages, direct testing is always best, but if the total amount of dispersent/surfactant in cement, and/or a concrete mix, can not be easily determined, then an indirect test method (like the Duggan test method or something similar) is absolutely required for the production of long-term, durable PCCP. How is a user of concrete to know, if too much surfactant/dispersent is in a mix? Perhaps the Duggan test method should be implemented during the time it takes to develope direct testing techniques that can qualify and quantify all of the components in cement and concrete.
In Iowa, approximately 10% of the cement produced goes towards highway and bridge construction. The cement companies are primarily in competition to supply ready-mix operations. The cement that can guarantee the highest early strength is the easiest to sell to a ready-mix owner. The ready-mix operator guarantees a certain early strength concrete. If he/she can get that strength using less cement (high cost concrete constituent), profits will increase. Since 90% percent of their product goes to ready-mixes and other concrete fabricators (non-highway), cement manufacturers are going to formulate their product towards optimum high early strength. Who knows what compounds are going into cement to attain high early strength? It has become more difficult, for DOT personnel, to obtain the raw products used in the manufacture of cement. Why should a cement company divulge its trade secrets? It can also be argued that the cement manufacturers are merely responding to external pressure from DOT and university concrete designers for high early strength cement formulations.
Some concrete investigators at the Iowa DOT suggested that new cement specifications should be implemented that would place lower limits on the amount of sulfur and potassium allowed in type I cement. At least one other state accomplished this goal by reducing the equivalent alkali in cement to 0.4 but not necessarily with the intent of limiting potassium. Limiting the amount of potassium in cement may be a step in the right direction but it would not always solve the probem. Cement potassium, tied to compounds other than arcanite (K2SO4) may not be a big problem just as magnesium tied to compounds other than periclase (MgO) is less of a problem. Also, even though the amount of grinding aid used should relate to the amount of arcanite in the cement, it does't have to be. If the amount of grinding aid, rather than the amount of arcanite (in the cement) is the culprit, then limiting the amount of potassium will not guarantee low Duggan test results.
The topic of arcanite in cement generated considerable discussion in the DOT PCCP durability meetings. Some cement company representatives said the melting point of arcanite precluded it from forming in cement clinkers generated at high temperatures. In theory, they should be correct. However, clinkers were dry sawed and dry polished (arcanite is highly soluble in water) and then examined in the SEM. Many clinkers, from some cement plants, were found to have a yellow center. What would normally be voids, in the center of these clinkers, was filled with arcanite. Small particles of arcanite can also be easily identified, in cement, using the element mapping facility of the SEM. Potassium and sulfur will overlay each other. The calcium/sulfur (gypsum, anhydrite or hemi-hydrite) overlay was not seen when examining the dry-polished clinkers.
It is not known why the arcanite can appear in the cement clinker. The recirculated, hot air, from the kiln, is used to preheat raw materials during cement production. This hot air can contain contain a significant amount of sulfur. If an overload of sulfur (in the recirculating air) occurs, it has to go someplace. Only enough fresh air is introduced into the system (for economic reasons) to maintain an oxidation mode (as opposed to reduction mode) of production.
Also, for economic and environmental reasons, high sulfur kiln dust is often returned to the final product (cement). The way cement is made now days is quite differnt than in the past. When early deteriorating concrete is investigated, all components in the mix should be suspect as well as the mix design. As far as blame is concerned, it is time to stop singling out the aggregate fraction of the mix.
For those PCCP mix designers who think that a slow gain in strength, over a period of years, relates to long-term PCCP service life, there are a couple of examples that could be followed. The New York water works department has cement (Merriman) made to their specifications. The state of South Dakota owns (or did at one time) a cement manufacturing plant. In either case, cement could be manufactured to user specifications. Inspectors could monitor the amount and type of all components used in the manufacture of the cement. Another technique that might work would be to mandate the use of easily detectable tracers in the concrete components that are difficult to measure by ordinary means. It is possible that the best cement formulations for ready-mix plants are not the best formulations for PCCP.
