Click for next page ( 50


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 49
49 Table 3.34. Resilient modulus and CBR Table 3.35. Water permeability of values for selected CLSM mixtures after selected CLSM mixtures. 28-day moist curinga. Permeability Mixture Mixture Regression Equation R2 CBR (%) (mm/s) I-1 2.46 10-3 Mr=3.001010 (Sd)-3.3517 I-2 5.33 10-4 Sd Total Mr I-3 1.45 10-4 4 (kPa) (GPa) 0.9155 215.93 I-4 4.20 10-3 276 199.45 I-5 6.75 10-3 345 66.23 I-6 2.89 10-4 414 52.52 Mr=3.00106 (Sd)-2.6284 Sd Total Mr (kPa) (GPa) was placed on this topic. A method developed in Germany to 6 0.8486 175.83 69 69.28 measure the shrinkage of conventional concrete for flooring 138 3.01 207 1.94 applications was used without modification in this study to 276 2.06 measure the shrinkage of CLSM mixtures. The results are Mr=1.46105 (Sd)-2.2978 shown in Table 3.37. The temperature during the testing pe- Sd Total Mr riod was 20C and the relative humidity was approximately (kPa) (GPa) 60 to 65 percent. For mixtures without air entrainment, most 23 34.5 57.61 0.9155 20.01 69 4.79 of the shrinkage occurred during the first day, probably due 103.5 3.49 to early bleeding and subsidence; however, the method used 138 2.36 in this study could not detect this shrinkage. It is possible that Mr=1.00108 (Sd)-2.8847 all of the shrinkage could not be detected because CLSM did Sd Total Mr (kPa) (GPa) not exhibit sufficient early strength (or stiffness) to cause a 24 138 134.32 0.9492 61.76 detectable movement of the end anchor. More research is 207 20.63 276 11.19 needed to examine drying shrinkage of CLSM and to develop 345 8.47 a suitable test method. Because the topic of drying shrinkage 414 4.74 was not identified as a critical issue for this project, no further Mr=3.06102 (Sd)-0.4929 research was performed on this topic. Sd Total Mr (kPa) (GPa) 207 23.33 Durability Test Methods 22r 276 17.99 0.9395 114.68 345 16.6 Corrosion 414 15.77 483 15.25 Phase I, Uncoupled Samples. To evaluate the potential 552 13.54 influence of resistivity, pH, fly ash type, fine aggregate type, Mr=6.57105 (Sd)-1.4393 watercementitious materials ratio (w/cm), and cement Sd Total Mr (kPa) (GPa) content on the corrosion activity of ductile iron coupons 26 0.8122 150.00 414 104.44 embedded completely in CLSM or sand, the percent mass 483 89.48 690 72.93 loss of coupons embedded in thirty different CLSM mixtures 828 33.25 (and eight duplicated mixtures) was evaluated. The box plot a Confining pressure 21 kPa. showing the distribution of the percent mass loss values of Sd = Deviator stress, Mr = Resilient modulus. the ductile iron coupons is given in Figure 3.13. Because mixtures 21 and 23 were not significantly different from other mixtures but the results obtained from them seem to increase in cohesion was the dominant factor. These mixtures be an anomaly, their data were not included in the statistical (I-4 and I-5) had high air contents and exhibited behavior analysis. similar to loose or uncompacted sands. It was interesting to A multiple regression analysis and an analysis of variance note that for mixture I-6, the friction angle decreased and co- were performed with the logarithm of percent mass loss hesion greatly increased with time. data of the 36 CLSM samples as the response variable. Com- parison of all possible main effect models for the maximum Drying Shrinkage adjusted R2 and minimum mean sum of error (MSE) indi- As stated earlier, there are no standard methods to evalu- cates that the best model to predict mass loss of ductile iron ate the drying shrinkage of CLSM and only limited emphasis pipe completely embedded in CLSM has three explanatory

OCR for page 49
50 Table 3.36. Results of triaxial shear tests. 7 days 28 days Mixture Friction angle ' () Cohesion c' (kPa) Friction angle ' () Cohesion c' (kPa) I-1 36.14 31.8 42.81 40.1 I-2 36.07 96.1 38.73 174.7 I-3 39.35 251.7 47.86 346.2 I-4 21.99 43.9 23.92 43.1 I-5 19.48 89.8 18.47 130.0 I-6 37.30 44.4 33.86 93.4 Table 3.37. Drying shrinkage of selected CLSM mixtures. Shrinkage Strain (x 10 -6 ) Time Mixture 4 Mixture 24 Mixture 23 Mixture 6 Mixture 22r Mixture 26 1 day 2260 2830 80 1440 90 10 2 days 2280 2850 80 1450 90 30 3 days 2280 2860 80 1450 90 50 4 days 2280 2860 80 1450 100 70 5 days 2300 2860 80 1460 100 100 6 days 2310 2880 80 1480 110 130 7 days 2330 2880 80 1500 120 160 2 weeks 2390 2930 160 1540 150 200 3 weeks 2410 2960 180 1590 150 190 4 weeks 2410 2980 180 1529 160 210 5 weeks 2410 2980 180 1600 160 220 6 weeks 2420 2960 180 1610 160 220 7 weeks 2410 2960 180 1600 " " = Not enough specimens were available for testing at this age. 5 Mixture 23 4 Mixture 21 Percent mass loss 3 2 1 0 All CLSM samples Figure 3.13. Box plot of percent mass loss values.

OCR for page 49
51 variables--fly ash type, fine aggregate type, and w/cm--as decrease in the percent mass loss as a result of the increased shown below: pH. As such, the pH of the pore solution alone does not seem to reliably estimate the corrosion performance of ductile iron w log10 ( % mass loss ) = 0.056 - - + 0.0312 (3.4) coupons embedded in CLSM. cm Statistical analysis of the data indicated that the mean log- arithm of percent mass loss values for mixtures containing where bottom ash, concrete sand, and foundry sand were statisti- = 1.13, 1.07, 1.31, and 0.0 for bottom ash, concrete sand, cally not different from each other. However, the mean loga- foundry sand, and no fine aggregate, respectively rithm of percent mass loss data for the coupons embedded in = 0.47, 0.61, 0.69, and 0.0 for Class C, Class F, high-carbon, mixtures without fine aggregates were statistically different and no fly ash, respectively. and higher than the other mixtures containing fine aggre- gates. The decrease in percent mass loss could be due to re- The logarithm of the percent mass loss data is the response ductions in the diffusivity, permeability, and/or porosity of variable. The adjusted RP2P for this second model is 67 per- the CLSM mixtures containing fine aggregates. cent and its MSE is 0.0916. Appropriate coefficients should be As noted, the amount of cement used in CLSM mixtures used to predict the expected mean percent mass loss for spe- is very low compared to the amount of water used. The sta- cific CLSM mixtures. The coefficients for the fine aggregate tistical analysis indicates that the cement content had no sig- type and the fly ash type are significant at the 95 percent con- nificant effect on the percent mass loss of the ductile iron fidence level and the coefficient for the w/cm is significant at coupons embedded in the CLSM mixtures. However, the re- the 89 percent confidence level. sults indicated a slight increase in the logarithm of percent Many field investigations on the corrosion of metals embed- mass loss values with increasing watercementitious materi- ded in soils have reported that resistivity is a major controlling als ratio. parameter affecting corrosion activity of the embedded metal (Spickelmire 2002, Kozhushner et al. 2001). Prior corrosion Phase I, Coupled Samples. To evaluate the mass loss research in soils reported a non-linear relationship between (i.e., corrosion performance) of the coupled ductile iron mass loss and resistivity (Edgar 1989, Palmer 1989). How- coupons embedded in both CLSM and sand, a similar statis- ever, the evidence for such a non-linear relationship for the tical analysis as described in the Phase I uncoupled samples CLSM data in this study is very weak. The sand used in study was performed. This analysis indicated that a good pre- the control samples exhibited a resistivity of 3.1 104 -cm diction of mass loss using the explanatory variables--cement and the average percent mass loss for the control group was content, fine aggregate type, fly ash type, etc.--was not pos- 0.39 percent. Ductile iron coupons embedded entirely in sible for ductile iron coupons embedded in two different en- CLSM exhibited lower corrosion activity than the ductile iron vironments (i.e., the coupled sample). coupons embedded in the control sand even though the re- The corrosion of uncoupled coupons was likely due to sistivity of the control sand material was higher than the re- the formation of micro-galvanic corrosion cells on the sur- sistivity of all the CLSM mixtures. This result is contradictory face of a single coupon. However, the major driving force of to conventional soil corrosion studies. the corrosion of ductile iron coupons coupled in two dif- Some utility agencies have voiced concern that the use of ferent environments was likely the formation of macro- fly ash in CLSM could be detrimental to the corrosion per- galvanic corrosion cells due to the potential difference formance of metals embedded in CLSM, because fly ash may between the ductile iron coupons. Because these macro- cause a reduction in the pH, which could further result in galvanic cells were the major driving force of the corrosion of higher corrosion activity. The results of this study indicate coupled coupons, factors that significantly affected the cor- that the logarithm of the mean percent mass loss of mixtures rosion of uncoupled coupons were insignificant for coupled without fly ash is statistically significantly higher than the coupons. mixtures with fly ash. This result indicates that the benefits of Figure 3.14 compares the logarithm of the distribution of the fly ash on the microstructure and long-term passivation percent mass loss of uncoupled coupons, coupled coupons, characteristics, as reported by Cao et al. (1994), likely have a and the control group. The distributions are grouped by fly more significant impact on corrosion performance than the ash type. The figure indicates that the coupling of the ductile relatively limited reduction in pH. iron coupons has a significant impact: the mass loss of duc- The mean pH of the pore solution from the CLSM mix- tile iron coupons embedded in sand and CLSM (i.e., coupled) tures evaluated in this study was 11.35. Although this high pH can be expected to be significantly larger than the mass loss of value was expected to decrease the corrosion activity of the the coupons completely embedded in CLSM and the control ductile iron coupons, the results do not indicate a significant group samples.

OCR for page 49
52 1.0 0.5 Log10 (Percent Mass Loss) 0.0 C: Type C Fly Ash F: Type F Fly Ash -0.5 HC: High Carbon Fly Ash None: No Fly Ash Control: Sand -1.0 -C: Indicates Coupled Samples -1.5 -2.0 -2.5 C-C C F-C F HC-C HC None-C None Control Figure 3.14. Uncoupled versus coupled log mass loss versus control group. In general, the results of the Phase I study indicated the ficient of multiple determination (R2) and root mean following: square values. Models were applied to the observed percent mass loss values, to their square root transformation, and to The corrosion activity for ductile iron pipe coupons com- their logarithm. Trials indicated that a logarithmic trans- pletely embedded in CLSM was significantly lower than formation was more effective in decreasing the observed de- that of ductile iron pipe embedded in sand. pendence of variability of residuals on the values of re- CLSM may provide more protection against corrosion sponse variable. Among the models evaluated for the initiation and propagation when metallic structures are logarithm of percent mass loss (LPML) values, the follow- completely embedded in CLSM compared to compacted ing model had the highest R2 value and smallest root means sand. square error: Examination of the effects of the constituent materials on corrosion with a limited number of samples indicated that log10 ( % mass loss ) = 1.844 + + + + ( + ) log 10 ( resistivity ) + pH + + ( + ) there was no significant difference between the fly ash types and the fine aggregate types used in this study. However, w the corrosion of metal coupons in uncoupled samples that + + + + (3.5) cm contained a fine aggregate or a fly ash was lower compared to the coupons in uncoupled samples without a fine aggregate or a fly ash. The model includes the following relationships: Phase II, Uncoupled Samples. Figure 3.15 shows the box The main effects of classification variables: environment (), plot showing the distribution and the median of the percent fine aggregate type (), fly ash type (), and metal type () mass loss data of the 361 galvanized steel and ductile iron The main effects of continuous variables: logarithm of resis- coupons embedded in CLSM mixtures exposed to distilled tivity (), pH (), and water-cementitious material (w/cm) water and chloride solution. ratio () A multiple regression analysis and analysis of variance The interaction effects of classification variables with clas- were performed on the data. The percent mass loss data sification variables: fine aggregate type with metal type (), were used as the response variable and the environment, fly ash type with metal type (), environment with metal fine aggregate type, fly ash type, resistivity, pH, metal type, type () and fly ash type with environment () watercementitious materials ratio, percent chloride con- The interaction effect of a classification variable with a con- tent, and cement content were used as the explanatory vari- tinuous variable: logarithm of electrical resistivity with ables. Different possible models consisting of main effects metal type () and w/cm with metal type (). and single interaction effects of the explanatory variables were applied to the data to find the best parsimonious model. However, further evaluation of the model indicated that Different models were compared using their adjusted coef- the assumptions of residuals being normally distributed

OCR for page 49
53 50 40 Percent Mass Loss 30 20 10 0 Uncoupled Samples Figure 3.15. Percent mass loss distribution of metallic coupons. and being independent of the predicted values of LPML means method at the mean values of three continuous vari- were not satisfied very well. Because the assumption of con- ables and at the 64 selected combinations of these three con- stant variance was not satisfied, a weighted regression analysis tinuous variables. Detailed information on the comparisons was performed. The factors that had the largest effect on the and the selection of combinations is given in Appendix A. LPML values were the environment and the metal type. The Analysis indicated that pH was significantly and in- variances of the four groups obtained by separating the versely correlated to the observed LPML values. Environ- data by environment and metal type were calculated. The ment was also a significant variable for all the samples. The reciprocals of variances of these four groups were used as a samples exposed to a chloride solution exhibited signifi- weight variable for the weighted regression analysis. Evalu- cantly higher LPML values compared to the samples ex- ation of the studentized residuals of the weighted regres- posed to the distilled water. The effect of environment for sion indicated that the normality assumption of residuals galvanized steel coupons was larger compared to the duc- was satisfied much better compared to the earlier regres- tile iron coupons. sion analysis. The R2 value for the weighted regression There was a significant difference in the LPML values of dif- analysis is 67 percent and the root mean square error value ferent metal types. For low watercementitious materials ra- is 0.98. All of the factors included in the model were statis- tios and logarithm of resistivity values, ductile iron coupons tically significant. exhibited significantly lower LPML values. However, at higher The parameters defined in the model for the main effects watercementitious materials ratios and with increasing log- of classification variables represent the expected value of the arithm of resistivity, the difference in values became smaller response variable for different levels of the corresponding and, at high enough values of these continuous variables, duc- classification variable, all other factors being the same. The tile iron coupons exhibited higher LPML values. parameters defined in the model for the main effects of con- The effects of different fly ash types and fine aggregate tinuous variables represent the amount of change in the ex- types were more important for samples with ductile iron pected value of the response variable for each unit change of coupons. Samples that contained a fine aggregate exhibited the corresponding continuous variable, all other factors being lower LPML values compared to the samples without fine ag- the same. The interaction parameters in the model define gregates regardless of the type of the fine aggregate. The dif- how the response reacts to one variable based on the value or ference between the mean LPML values of samples contain- level of another variable. In the case of an interaction of a clas- ing bottom ash and sand as fine aggregates was statistically sification variable with a continuous variable, the coefficient not significant. The samples containing foundry sand as fine of the continuous variable is changed based on the level of the aggregate exhibited a mean LPML value between that of the classification variable. The values of the parameters are given samples with bottom ash or sand and the samples without in Appendix A. fine aggregates. Because of the high LPML variability of sam- In addition to the regression and variance analyses, com- ples containing foundry sand, the difference between these parisons of LPML values for the different levels of classifica- samples and the samples without fine aggregates was not sta- tion variables were performed using Tukey's comparison of tistically significant.

OCR for page 49
54 Based on the materials used for this study, the results, from embedded in the soil sections of coupled samples represent this phase only, indicate that the use of fly ash as a supple- the critical anodic areas of pipes for corrosion damage, fur- mentary cementitious material may have adverse effects on ther statistical analysis was performed on the percent mass the corrosion of embedded galvanized steel or ductile iron loss data of these coupons. coupons, especially for the ductile iron coupons. Samples The explanatory variables evaluated for the percent mass containing high-carbon fly ash or Class F fly ash exhibited loss of coupons included environment, metal type, soil type, higher LPML values compared to the samples without fly fine aggregate type, fly ash type, resistivity of CLSM, resis- ashes, but the samples without fly ashes exhibited much tivity of soil, pH of CLSM, chloride content of the CLSM, larger variation. The mean LPML value of the samples con- and chloride content of the soil. Different possible models taining Class C fly ash was lower than that of the samples with consisting of main effects and single interaction effects of Class F or high-carbon fly ash but higher than that of the sam- the explanatory variables were applied to the data to find the ples without fly ash. However, due to the high variance of the best parsimonious model. Different models were compared samples without fly ash, the difference between the samples using their adjusted coefficient of multiple determination containing Class C fly ash and samples without fly ash was not (R2) and root mean square values. Models were applied to statistically significant. the observed percent mass loss values, to their square root transformation, and to their logarithm. Trials indicated that Phase II, Coupled Samples. The histogram showing the a logarithmic transformation was more effective in decreas- percent mass loss of ductile iron and galvanized steel coupons ing the observed dependence of variability of residuals on embedded in CLSM and soil sections of coupled samples ex- the values of response variable. Among the models evalu- posed to distilled water and chloride solution are shown in ated for the logarithm of percent mass loss values, the fol- lowing model had the highest R2 value and smallest root Figure 3.16. mean square error: Analyses indicate that the percent mass loss values of metallic coupons embedded in CLSM and soil were signifi- log10 ( % mass loss ) = 0.97 + + + + + + + cantly correlated and the mass loss values of coupons em- bedded in the soil section of samples were higher compared + + (3.6) to the mass loss values of coupons embedded in the CLSM section of samples. For the coupled coupons, the mass loss is The coefficients , , , , and are assigned values for the believed to be mainly due to galvanic corrosion taking place different levels of the classification variables: environment between the metallic coupons embedded in different sections. (), soil type (), fine aggregate type (), fly ash type (), and The significantly higher mean percent mass loss values ex- metal type (), respectively. The coefficients , , , are as- hibited by the metallic coupons in the soil section indicate signed values for the two-factor interactions of classification that these coupons were anodes and the coupons in the variables: environment with metal type (), environment CLSM section were cathodes. Because the metallic coupons with soil type (), fly ash type with metal type (), and fine 600 500 400 Frequency 300 200 100 0 0 12 24 36 48 60 72 84 Percent Mass Loss Figure 3.16. Percent mass loss of metallic coupons.

OCR for page 49
55 aggregate type with soil type (). The values of these coeffi- In general, results from the Phase II study indicated the cients are given in Appendix A. following: Analysis showed that the overall model and all the factors included in the model were significant with the exception of pH was significantly and inversely correlated to the ob- metal type. However, because interactions of other variables served LPML values. with metal type were significant, this factor was left in the Environment was a significant variable for all the samples. model for a complete hierarchy. All assumptions of the re- The samples exposed to a chloride solution exhibited sig- gression analysis were satisfied and the coefficient of deter- nificantly higher LPML values compared to the samples mination (R2) of the model was 35 percent. This low R2 value exposed to the distilled water. indicates that a model solely built from these variables can- The effect of environment for galvanized steel coupons was not be used to estimate the corrosion of metallic coupons larger compared to the ductile iron coupons. with great accuracy. There was a significant difference in the LPML values of Results indicated that samples exposed to chloride environ- different metal types. ment exhibited higher mean LPML values compared to the For low w/cm and logarithm of resistivity values, ductile samples exposed to distilled water. The disturbance of the pas- iron coupons exhibited significantly lower LPML values. sive layer formation on the steel surface by chloride ions and At higher w/cm and with increasing logarithm of resistivity, the low resistivity of CLSM and soil samples exposed to chlo- the difference in values became less and, at sufficiently high ride solution could both be the reasons for the higher mean values, ductile iron coupons exhibited higher LPML values. LPML values of samples exposed to chloride environment. The effects of different fly ash types and fine aggregate types Watercementitious materials ratio had a statistically sig- were more important for samples with ductile iron coupons. Samples that contained a fine aggregate exhibited lower nificant but small correlation with the chloride content in CLSM and a negative correlation with the logarithm of resis- LPML values compared to the samples without fine aggre- tivity of CLSM. gates regardless of the type of the fine aggregate. The difference between the mean LPML values of samples Results indicated that coupons embedded in clay (soil sec- containing bottom ash and sand as fine aggregates was sta- tion of coupled samples) exhibited statistically significantly tistically not significant. higher LPML values compared to the coupons embedded in The samples containing foundry sand as fine aggregate ex- sand in both chloride and distilled water environments. How- hibited a mean LPML value between the samples with bot- ever, the effect of environment was greater on the coupons tom ash or sand and the samples without fine aggregates. that were embedded in sand compared to the coupons em- Because of the high LPML variability of samples contain- bedded in clay. Analysis also indicated that the resistivity and ing foundry sand, the difference between these samples pH of clay samples were lower compared to the resistivity and and the samples without fine aggregates was not statisti- pH of sand. cally significant. Although the metal type was overall not a statistically sig- The use of fly ashes may have adverse effects on the corro- nificant factor, analysis indicated that galvanized steel cou- sion of embedded galvanized steel or ductile iron coupons, pons exhibited a significantly higher mean LPML value com- especially for the ductile iron coupons. pared to ductile iron coupons in a chloride environment. Samples containing a high-carbon fly ash or Class F fly ash The observed effect of fly ash on the LPML was contradic- exhibited higher LPML values compared to the samples tory to the findings of the uncoupled samples. Results indi- without fly ashes, but the samples without fly ashes exhib- cated that, among the coupled samples, CLSM sections with ited much larger variation. fly ashes exhibited lower mean LPML values compared to the The mean LPML value of the samples containing Class C fly CLSM sections without any type of fly ash (similar to the ash was lower than that of the samples with Class F or high- Phase I study). However, the difference between the LPML carbon fly ash but higher than that of the samples without values of CLSM sections containing fly ash and without fly ash fly ash. However, because of the high variance of the sam- was only statistically significant for the ductile iron coupons. ples without fly ash, the difference between the samples con- CLSM sections with bottom ash or foundry sand exhibited taining Class C fly ash and samples without fly ash was not significantly higher LPML values compared to the CLSM sec- statistically significant. tions with sand or without fine aggregates. Among the cou- pled samples that had clay in their soil section, CLSM sections The following general conclusions were obtained from with bottom ash exhibited the highest mean LPML value and both phases of the study: among the coupled samples that had sand in their soil sec- tion, CLSM sections with foundry sand exhibited the highest The metallic coupons embedded in the soil section of cou- mean LPML value. pled samples exhibited significantly higher percent mass

OCR for page 49
56 loss values compared to the coupons embedded in uncou- tain a distribution around the obtained service life value. pled samples. Equation 3.8 shows how to obtain the required percentile of Because the main driving force of corrosion is the poten- the LPML value using the variance and the LPML value ob- tial difference in the coupled samples, the significance of tained from Equation 3.5: the factors that affected the corrosion in uncoupled sam- ples was generally lower for coupled samples. LPMLPr. = LPML + -1 ( Pr.) Variance (3.8) Service Life of Ductile Iron and Galvanized Steel where Coupons Completely Embedded in CLSM LPMLPr. = the LPML for which probability of LPML < LPMLPr. is Pr. ASTM G 1 provides a formula to predict the corrosion -1 = the inverse standard normal distribution rate of metallic samples. By placing the LPML values ob- function tained from the statistical model shown in Equation 3.5 into the formula given in ASTM G 1, a service life model for duc- After the values of the classification variables for a specific tile iron and galvanized steel pipes completely embedded in CLSM design are determined and a combination of the levels CLSM can be derived. Assuming that the useful service life of the pipe will be over at the first perforation of the pipe of continuous variables is chosen, a service life distribution wall due to corrosion, the service life of a pipe completely graph can be generated for a ductile iron or galvanized steel embedded in CLSM can be calculated using the following pipe completely embedded in the specific CLSM mixture as formula: shown in Figure 3.17. Calculation of service life estimates for the galvanized and ductile iron pipes embedded in the specific CLSM mix- D t 7978 x10-2 tures that were evaluated in this study indicated that prop- SL = (3.7) 10 LPML() D - (D - t ) 2 2 erly designed CLSM mixtures can provide a service life for ductile iron pipes similar to that in conventional backfill where materials. Therefore, in selecting between CLSM and SL = the service life (years) conventional backfill materials, factors other than service D= the outside radius (cm) life--such as material cost, construction cost, construction t= the pipe wall thickness (cm) time, and long-term settlement--should be considered. LPML() = the logarithm of percent mass loss obtained However, results indicated that the galvanized steel pipes from Equation. 3.5 completely embedded in CLSM can be expected to have service life values comparable to the galvanized steel pipes To obtain the LPML value from Equation 3.5, the values embedded in severely or moderately corrosive soils with of the classification variables and the values of the three low resistivity and pH values. Therefore, backfilling bare continuous variables (w/cm, resistivity, and pH) must be galvanized steel pipes with CLSM mixtures is likely not specified. However, only the values of the classification warranted. variables, such as fly ash type, fine aggregate type, environ- ment, etc., can be specified. The values of the continuous Freezing and Thawing variables are dependent values, i.e., they cannot be speci- fied; they can only be measured from the samples of de- Two studies were performed to evaluate the freeze-thaw signed CLSM mixtures. Therefore, Equation 3.7 cannot be resistance of CLSM mixtures. In the first study, six CLSM used to calculate a specific service life for a designed CLSM mixtures from the initial mixture series were tested using mixture. However, the formula can be used to perform a a modified version of ASTM D 560, originally developed risk analysis by using different combinations of the contin- for the assessment of soil-cement. The second study used uous variables in the LPML formula. The data obtained in the same method to assess a wider range of CLSM mixtures this study can be used to obtain an estimate of the expected (D-series) and evaluated the effects of freeze-thaw damage on range of the continuous variables for different levels of the permeability. classification variables. Figure 3.18 shows the measured percent mass loss values It should also be noted that the coefficients of the LPML plotted against the number of freeze-thaw cycles. The per- model were determined using a weighted regression analysis. cent mass loss values shown in the figure were calculated as- The variance of each residual group that was used to deter- suming that the moisture content of all the specimens were mine the weight variables of the analysis can be used to ob- constant throughout the test. According to the "soil-cement

OCR for page 49
57 100 80 Pr (Servicelife < X) 60 40 20 0 0 50 100 150 200 250 Service Life (years) Figure 3.17. Probability distribution of service life. laboratory handbook" the mass loss after 12 freeze-thaw cy- that mixture D-1 (high-air mixture with 30 kg/m3 of cles should not exceed 14 percent for a Group A-1 soil (PCA cement) did not survive the 12 cycles, whereas a similar 1992). Soils in group A-1 are coarse grained and low in con- mixture with 45 kg/m3 of cement (D-11) did survive the en- tent of fines. Most of the CLSM mixtures (especially the tire 12 cycles, suggesting that both air-void system and ones that were cured for 28 days) containing high amounts strength contribute to freezing and thawing resistance. of air satisfied this criterion. ASTM D 560 (with minor Mixture D-10 survived all 12 cycles, most likely due to its modifications) was found to be an effective and easy-to- higher strength (contributed from the Class C fly ash). The perform method to assess the freeze-thaw resistance of CLSM remaining mixtures (D-2 through D-9) did not survive all mixtures. 12 cycles. Mixtures were selected that would likely suffer The permeability or hydraulic conductivity of the D- freezing and thawing damage, allowing for the measure- series CLSM mixtures before and after exposure to freeze- ment of changes in permeability (before and after testing, thaw cycles is shown in Table 3.38. It is interesting to note as shown in Table 3.38). However, the effects of freezing 60 #4 cured for 7 days #23 cured for 28 days 50 #6 cured for 7 days #6 cured for 28 days #22r cured for 7 days #22r cured for 28 days 40 #26 cured for 7 days #25 cured for 28 days #4 cured for 28 days Mass Loss (%) 30 20 10 0 -10 0 2 4 6 8 10 12 14 Number of Cycles Figure 3.18. Mass losses vs. freeze-thaw cycles for selected CLSM mixtures.

OCR for page 49
58 Table 3.38. Frost resistance of CLSM (using modified ASTM D 560). Mixture Measurement D-1 D-2 D-3 D-4 D-5 D-6 D-7 D-8 D-9 D-10 D-11 Original mass (kg) 1.38 1.77 1.70 1.86 1.53 1.47 1.46 1.50 1.55 1.56 1.33 Moisture content (%) 9.8 10.5 13.6 10.1 30.7 35.5 32.4 27.4 28 25.1 7.3 28-day strength (MPa) 0.13 1.02 0.79 1.47 0.11 0.12 0.20 0.38 0.25 0.53 0.34 1 cycle (%) 105.2 99.0 100.1 99.9 D 92.8 93.0 98.9 98.1 99.2 103.5 2 cycle (%) 102.3 95.3 97.3 99.8 D D 94.9 94.8 98.1 105.9 3 cycle (%) 84.4 85.4 78.4 99.1 91.5 87.1 98.0 108.1 4 cycle (%) 74.1 75.7 73.2 96.4 88.0 83.6 98.1 110.3 5 cycle (%) 66.4 72.4 71.1 88.4 84.6 79.1 97.7 110.3 6 cycle (%) 59.6 55.4 63.8 61.0 81.4 73.5 97.3 110.3 7 cycle (%) 51.5 54.9 52.1 59.7 76.9 69.2 97.1 108.8 8 cycle (%) 45.8 46.3 54.3 57.9 73.7 62.1 95.8 108.2 9 cycle (%) 37.3 30.1 48.6 43.8 68.1 56.3 94.4 104.6 10 cycle (%) 33.6 D D 40.3 56.1 D 92.2 103.6 11 cycle (%) 25.8 D 49.8 89.4 100.6 12 cycle (%) D D 85.1 97.5 Final moisture content 28.7 21.3 (%) Dry mass loss (%) 11.0 11.9 Permeability before 12 F-T cycles 7.38 0.35 0.21 13.02 6.60 3.87 1.63 1.60 3.08 3.63 8.06 ( 10-2 mm/s) Permeability after 12 F-T cycles 14.21 1.90 1.61 8.94 0.08 0.94 1.72 0.42 0.73 0.36 5.94 ( 10-2 mm/s) D = Damaged. and thawing damage on water permeability were somewhat Leaching and Environmental Impact inconclusive, with some mixtures showing increased per- Table 3.39 summarizes the total concentration of heavy meability and others showing decreased permeability. This metals present in the by-product materials used in this result was most likely due to the test setup, which was study. These results represent the total concentration of the designed to keep the samples intact, thus allowing for sub- eight key heavy metals. A "rule of thumb" that some practi- sequent permeability testing. However, keeping the sam- tioners use is that the concentration of total heavy metals ples intact (and confined) may not have allowed for an ac- can be up to 20 times the standard TCLP limits. In this curate estimate of in-situ permeability. Because the samples study, arsenic concentration in bottom ash, Class C fly ash, were confined, the expansion due to freezing and thawing and Class F fly ash exceeded this "rule of thumb" value. may have actually compacted the samples, resulting in an Thus, additional testing was performed (using the TCLP apparent reduction in permeability. More work is needed method) to determine the actual amount of heavy metals to elucidate the effects of freezing and thawing damage that are available to leach from these materials. Because the on permeability. Initially, the composition of the water foundry sand and high-carbon fly ash did not have signifi- flowing through the sample was to be analyzed to deter- cant amounts of total heavy metals, the materials were clas- mine if freezing and thawing damage increased the leach- sified as non-toxic, and no subsequent leaching tests were ing of constituent materials, specifically heavy metals. performed. However, after analyzing the raw materials used in the The TCLP results for Class C fly ash, Class F fly ash, and study (as discussed in the next section), the researchers de- bottom ash are shown in Table 3.40. The concentration of termined that the materials used were intrinsically non- heavy metals that leached from each material was well toxic. Thus, the effluent from the freeze-thaw samples was below the EPA-recommended TCLP limits; therefore, not analyzed. the materials were classified as non-toxic and suitable for In general, the results of the freeze-thaw testing indicated use in CLSM. If any of the by-product materials had exhib- that CLSM mixtures can be efficiently tested for freeze-thaw ited significant leaching of heavy metals (above the TCLP resistance following the modified ASTM D 560 with 12 cycles. limits), the last step would have been to assess the actual Results also indicated that CLSM mixtures with high air con- leaching of heavy metals from CLSM containing the mate- tent and high compressive strength exhibited good freeze- rial(s) using the American Nuclear Society leachate test thaw resistance. (ANS 16.1).