5££*.% *+..'J&i.S. c"*.c^:.* o >\. V* V W 1 .*tfta W .vriMfc V> v> A v ,0 *, ^ °^ * • - • VV -j»V ' V* '•*• • "° A v ' ** v % ^d* 'bV" ,v^ > ^ lV>^ i^ . K • - *rt .V 0* • , 4 O ^°* IC 8941 Bureau of Mines Information Circular/1983 Field Determinations of a Probabilistic Density Function for Slope Stability Analysis of Tailings Embankments By P. C. McWilliams and D. R. Tesarik UNITED STATES DEPARTMENT OF THE INTERIOR Information Circular 8941 1 Field Determinations of a Probabilistic Density Function for Slope Stability Analysis of Tailings Embankments By P. C. McWilliams and D. R. Tesarik UNITED STATES DEPARTMENT OF THE INTERIOR James G. Watt, Secretary BUREAU OF MINES Robert C. Horton, Director A 0' This publication has been cataloged as follows: McWilliams, P. C. (Paul C.) Field determinations of a probabilistic density function for slope stability analysis of tailings embankments. (Information circular / Bureau of Mines ; 8941) Bibliography: p. 13. Supt. of Docs, no.: I 28.27:8941. 1. Spoil banks— Safety measures. 2. Tailings dams— Safety mea- sures. 3. Slopes (Soil mechanics). 4. Soil mechanics— Statistical methods. 5. Probabilities. I. Tesarik, D. R. (Douglas R.). II. Title. III. Series: Information circular (United States. Bureau of Mines) ; 8941. -W295.U4 622s [622.7] 83-600123 CONTENTS Page Abstract 1 Introduction 2 Acknowledgments 2 Probabilistic approach 2 Propagation of error 3 Distribution properties 5 Propagation of error applied to the data 9 Confidence interval for the factor of safety 9 X Geotechnical innovation and probabilistic modeling 10 Conclusions 12 References 13 Appendix. — Data collection, slope stability, and laboratory testing 14 ILLUSTRATIONS a! \l 1. Slip circle factor of safety formulation 2 2. Concept of probability of failure 3 3. Factor of safety histograms 7 4. Internal friction angle histograms 7 5. Cohesion histograms 7 6. Soil density histograms 7 7. Ninety-five percent confidence interval — factor of safety, CDA Mine — with varying sample size 11 8. Ninety-five percent confidence interval — factor of safety, SW Mine — with varying sample size 11 9. Variance reduction applied to spatially oriented shear strength values... 12 A-l. SW data station pattern 15 A-2. Idealized profile — SW embankment 16 A-3. Range of grain-size curves for CDA data 17 TABLES 1. Basic statistics for soil parameters 4 2. Three model curve fits to Fellenius 1 factor of safety 6 3. Normal curve-fit parameters for ^ Si N s «\| w. FIGURE 1. - Slip circle \ formulation . KEY th = Weight of i ln slice = Normal force, i ,h slice = Resisting shear force Safety." A necessary ingredient in most probabilistic work is the assumption of a PDF that characterizes a population of factor of safety values. Another approach is to compute the probability of failure, which is repre- sented by the area under the PDF to the left of the value 1 (fig. 2). The main objective of this report was to collect enough sample values from two tailings embankments to establish the PDF for the factor of safety values. In practice, it would not be expedient to expend the time and money to collect the necessary number of samples required to derive the PDF. Work such as that done herein should pro- vide some assurance that it would not be unreasonable to postulate a particular PDF without actually collecting a partic- ularly large sample of data. PROPAGATION OF ERROR 4 An interesting alternative to collect- ing enough samples to establish a PDF is to apply the statistical technique known as propagation of error (6). This technique will approximate the standard deviation of the factor of safety (sp). It may not be convenient for the designer to collect enough data to compute Sp di- rectly; he or she may instead assume values for the coefficient of variation for the soil properties — and then apply propagation of error. Reference is invited to either Lee and Singh ( 11 ) or table 1 of this paper for candidate co- efficient of variation values. Mathematical follows: considerations are as Given a function of several variables, G = G (a, b, c), (1) 4 There between exist the notational disciplines ambiguities of the mathematicians-statisticians and that of the geotechnical engineers, per, deference is given to In this pa- the notation of mathematics and statistics. KEY F = Factor of safety F = Mean factor of safety s F = Standard deviation /\ of F A l F c = Critical factor Probability / ! / \ of safety of fa i i 1 u r e / ^s F -^ / Distribution of F >v F c = 1 . F FIGURE 2. - Concept of probability of failure, by propagation of error, the variance of G (s G ), is given by Sg = G a s a + G b s b + G c s c + 2r ab G a G b s a s b + 2r ac G a G c s a s c + 2r bc G b G c s c s b , where G\ = partial derivative of G with respect to the i"^ parameter, (2) and S| = variance of the i— parameter, r|j = correlation coefficient for i— and j— parameters Assume that the embankment of consideration is not "zoned"; that is, it is of a homogeneous nature. Fellenius' equation for the factor of safety (F) for soils may be written as - cL + tan = angle of internal friction. c = cohesion. Y = soil density. F = factor of safety. Mean = sample average value. S = sample standard deviation. Cw = coefficient of variation = 100-S ■ mean See figure 1 for the geometry of Fellenius' method. Applying the propagation of error formula to equation 3 yields ,2 2 s F = *> s^ + F c s c + F y s y + I F U| s u 2 r c ,y F c F Y s c s Y + . + 2r u n-1 + 2r c,<|> F c F <|> s c s "n-r u n &u n-r u n. (4) The required partial derivatives are u ? and Y E A| sin G| ' sec 2 <{> Z (y A| cos 0|- U| &l\) ~~y £ A| sin -tan <}> A£j Y E A| sin Oj ' -cL + tan has less variation than anticipated, although Singh's work focused on variability between laborato- ries rather than on within-laboratory variability. Note the close relationship between the values of Bishop's factor of safety and that of Fellenius. Since the propagation of error work uses Fellenius' factor of safety, we will only use Felle- nius' numbers henceforth. To better visualize the data, histo- grams of the data are shown in figures 3 through 6. In figure 3, histogram of the factor of safety, note the wide variabil- ity in the CDA data. Over 10 pet of the data lies to the left of 1; thus the probability of failure is predicted to be quite high. However, there are two points to keep in mind: paper considers the normal, log-normal, and Weibull distributions. Table 2 sum- marizes curve-fitting these functions to Fellenius' factor of safety values. Two interesting points from table 2: may be drawn 1 . The logarithmic transformation does not improve fitting the PDF's. This is reasonable, for the log-normal transfor- mation works best for data that are ini- tially skewed to the left. 1. In this case, the material is con- tained behind a berm of borrow material; thus the factor of safety values are sim- ulated, not actual values. 2. This wide variability gives credi- bility to the variance reduction concept, which is intrinsic to the probabilistic work of Vanmarcke and Anderson (2^, 16 ) . Thus, variance reduction, although a con- troversial concept as a "pragmatic value" at this time, is conceptually acceptable and at least partially verified as to need by this distribution. Certainly the histogram for the factor of safety (fig. 3) demonstrates a strong central tendency in both cases; the next logical question is whether a conven- tional mathematical PDF "fits" the data properly or not. There are many candi- date functions for consideration; this 2. The Weibull PDF does not improve on the curve fits for the normal PDF. Since the Weibull is more "obscure" (for exam- ple, parameters are not easily under- stood, and the curve-fitting is appreci- ably more difficult) , there seems to be no reason to use it in preference to the normal PDF. It is true that there are other candi- date distributions — the beta, gamma, Poisson, et al. — that could have been considered. In particular, the geotech- nical community has shown some preference for the beta PDF. Upon investigation of the properties of the Weibull and beta PDF's, it is found that both are versa- tile functions that can fit a variety of situations in a very similar manner. Our choice is to investigate just the Weibull PDF in this report. It is felt that this investigation is sufficient and the TABLE 2. - Three model curve fits to Fellenius' factor of safety Probability density functions 1 Index of determination 2 X 2 values 3 Probability density functions 1 Index of determination 2 X 2 values 3 CDA SITE SW SITE Log normal. . . 0.62 .59 .62 4 18.41 NA NA Log normal . . . 0.96 .94 .96 5 6.13 NA NA NA Not available. 1 A11 curve fitting done using Gauss' nonlinear iterative technique (13). Measures how well the data agree with curve-fit function. Reduces to the correla- used for comparison is 9.49 for these tion coefficient in the linear case (13) . 3 Goodness-of-fit test; the critical value cases. 4 Reject. 5 Accept. 20 > o Z 40 M O 3d Coeur d' Alene 20 ■ : Southwest F = 1.95 S c =0.24 0.6 1.0 1.4 1.8 2.2 2.6 3.0 F (FELLENIUS) FIGURE 3. - Factor of safety histograms. I I I ^^r\ Coeur flT Alene 10 c =3.53 S c =2.94 H i Vr I 40 > O z M O 30 LU X 20 10 COHESION, psi FIGURE 5. - Cohesion histograms. 30 20 10 40 30 20 10 25 30 35 40 45 50 PHI ANGLE, deg FIGURE 4. - Internal friction angle histograms. 20 uj ° 30 Coeur d' Alene Southwest FIGURE 6. - Soil density histograms. normal PDF is as good a choice as can be made for these two examples. It is true that the normal distribution's curve fit to the CDA is not outstanding as indi- cated by the chi-square measure. The data are simply not conducive to a good curve fit, regardless of what function one might try. Table 3 enumerates pertinent statistics and curve-fit parameters for internal angle of friction, cohesion, and density. In view of the preceding discussion, and since the remaining PDF's are of lesser interest to us, only the normal curve will be used to fit these histo- grams. The internal angle of friction distributions (fig. 4), although again demonstrating central tendency, are each skewed, both to the right. The index of determination for both fits is satisfac- tory, with a somewhat better fit for the SW data. Note that the fitted parameters vary somewhat from the sample means and standard deviations; the degree of vari- ability is a function of the goodness of fit. It is not surprising that the cohesion value (fig. 5) has the largest variabil- ity, for this is apparently inherent in the cohesion parameter ( 11 ) . Thus the normal curve approximations to the cohe- sion histograms are not particularly im- pressive, as indicated by the relatively low indices of determination. Here the histograms would indicate a slight skew- ness to the left. Finally, consider the histogram (fig. 6) for the soil density. Again, curve fitting these curves is of somewhat nebu- lous value; for example, a uniform dis- tribution may be a better candidate for the SW data. But to be consistent, the normal curve approximation was used; the resulting indices of dispersion reflect the degree of disagreement between the data and fitted function. To summarize, central tendency is ex- hibited by all of the histograms (and corresponding PDF's) considered. The normal curve seems to be an acceptable candidate model. For consistency's sake, the normal curve was then used for all TABLE 3. - Normal curve-fit parameters for , c, y, and F Normal curve-fit parameter Area Mean Index of determination 2 CDA SITE ♦ c Y F-Fellenius .deg.. .psi. . .pcf. . 118.78 115.22 147.38 14.14 0.82 .79 .62 .62 SW SITE * : c Y F-Fellenius, ,deg.. .psi. . .pcf. . 291.09 296.44 441.30 17.72 .91 .85 .88 .96 1 Equation for normal PDF used: - (x - mean) 2 /2 (std. dev.) 2 y = (area) /2rr (std. dev.) where y = dependent variable (histogram value) , x = independent variable (F, <)) , c, y)> 2 A measure of goodness of fit for nonlinear functions; see reference 13 for details. histogram curve fitting. Also, the se- verity of concern decreases considerably as one moves from the factor of safe- ty to the supporting data sets. Thus, for probabilistic work, in finding the probability of failure and in setting confidence intervals on the estimated mean factor of safety value, these two examples fortify use of the normal proba- bility density function. PROPAGATION OF ERROR APPLIED TO THE DATA As stated previously, it is desirable to take only a few field samples and use the propagation of error technique (6) to estimate the variance of the factor of safety. However, since we have complete sets of data available, we naturally used all these data for comparative slope sta- bility work herein. (See appendix for the data description.) Table 4 shows the coefficient of variation for Fellenius' factor of safety value with assumed pore pressure coefficient of variation values of 0, 10, and 20 pet. These values are compared to the sample coefficient of variation value. In the case of the SW embankment , the agreement between the propagation values and that of the sample data is quite good. The CDA data do not agree as well, but the divergence would seem acceptable. Note that the pairwise correlations between the soil parameters (for example, "" and "c") may or may not be included in the computations, depend- ing on the user's preference. We chose to use these correlation coefficients in our computations. Table 5 enumerates the pairwise correlation coefficients used in calculating table 4. TABLE 4. - Comparison of coefficient of variation estimates for Felle- nius' factor of safety Coefficient of variation pet 1 10 pet 1 20 pet 1 Actual sample values 2 1 The entries are the propagation of er- ror computed values for the specified pure pressure. 2 The entries are the comparative values obtained from the available sample data — see table 1. TABLE 5. - Correlation coefficients Correlation coefficient r c,4> r c,Y r ,Y CONFIDENCE INTERVAL FOR THE FACTOR OF SAFETY There is one basic statistical approach that can be easily applied to the factor of safety computation which provides a simplistic marriage between the world of probability and the presently used deter- ministic computation — putting a confi- dence limit about the factor of safe- ty value. A brief review of statistical principles would seem in order. Suppose one is concerned with a variable whose PDF is not necessarily normally distri- buted, but for which the population mean and standard deviation exist. Next, a sample size "n" is chosen. By some ran- dom process, "n" samples are selected from the original population and a sample mean (x) is computed. This process is then repeated until a population of sample means (all of which are of the preselected sample size "n") exists. The Central Limit Theory of Statistics states that the PDF of mean values will be nor- mally distributed provided "n" is suffi- ciently large. Furthermore, the popula- tion mean of the new distribution — the distribution of sample means — is the same as the population mean of the original PDF, but the standard deviation of the new distribution equals the original PDF's standard deviation divided by the square root of "n." When the Central Limit Theorem is actually applied to a data set, only one sample of size "n" is taken from which the sample mean and standard deviation are computed. 10 This powerful theorem can be applied to the distribution of factor of safety val- ues. The key of the theorem is, of course, how many samples are required for it to be valid. The answer to this query is certainly not simple; if the original distribution is very similar to the nor- mal curve, then small sample sizes suf- fice. For symmetrical distributions (the preceding work and that of Baecher and Marr (_3) substantiate that one may assume symmetry) , a sample size of eight or greater would seem appropriate. Thus the computational procedure would be as follows: 1. Collect a representative set of soil properties; a minimum of eight sets is recommended. Compute a corresponding set of factor of safety values. 2. Compute the sample mean (F) and standard deviation (s F ) for these factor of safety values. 3. Apply the Central Limit Theorem by forming a 95-pct confidence interval for the true but unknown population mean fac- tor of safety value (up). The following computation is requisite: where F ± to. 975 ( n ~l) * s f/ ^i = 95-pct confidence interval, n = sample size, and tg. 975 (n-1) = "t" distribution value, which depends on sample size and the level of confidence. - 1 There is no set rule that mandates a 95- pct confidence interval; this is a con- ventional and conservative choice. Let us investigate what it meant by a confidence interval. For example, a 95- pct confidence interval implies that if one were to form 100 such intervals , 95 of these would contain the true (but unknown) population parameter. In prac- tice, one computes only one such interval and relies on statistical theory to pro- vide the statement of confidence or assurance. A simulation may prove illustrative. The authors selected, via random sam- pling, 16 sample factor of safety values from each of the previously discussed data populations. It was assumed that the computed sample mean value was a best approximation to the true factor of safe- ty value. Then confidence intervals for subsamples of size 4 were formed, then of size 8, and finally of size 16. These confidence intervals are shown in figures 7 and 8 for the two data populations of interest. Note that in all but one case the 95-pct confidence intervals did con- tain the overall sample mean. Also, note that the intervals tend to lessen in length as the sample size increases. In figure 7 — the CDA data — samples of size 4 include factor of safety values less than 1 in three of the four cases. Also, one of the two samples of size 8 extends be- low a factor of safety value of 1. This is, of course, an indicator of concern for the embankment designer. Again, the preceding simulation is tu- torial, for in practice one would obtain only one confidence interval and use it to make a statement about the factor of safety. Furthermore, an example is, at best, illustrative and certainly does not replace theoretical considerations. Ideally, the interval would contain no factor of safety values below the criti- cal value of 1. GEOTECHNICAL INNOVATION AND PROBABILISTIC MODELING There is a new approach regarding the "best" model for the factor of safety calculation. The traditional two-dimensional slip circle of Fellenius and Bishop is being challenged. In par- ticular, a three-dimensional cylinder has 11 3.0 2.6 2.2 n = 4 >- LU < 1.8 — CO O 1.4 = O < 1.0 .6 (— n = 8 n = 16 F F = 1.43 , F I Accep Reject F FIGURE 7. - Ninety-five percent confidence interval-factor of safety, CDA Mine— with vary- ing sample size. been postulated by Vanmarcke and others (—> 16) • Theoretical development reduces the cylindrical model effectively to that of Bishop (5, 9_) , but with end-effect forces considered. A most important concept in the geo- technical works of Vanmarcke (16) and Anderson (2^) is variance reduction. This rather complex topic is explained in depth in the two aforementioned refer- ences; the following is a heuristic over- view. Rather than being interested in the usual statistical value of the vari- ance of the strength parameter, the geo- technical engineer is concerned when a succession of weak strength values oc- curs. In such a case, the possibility of failure is greatest. If one were to envision a moving average curve being fit to a spatial succession of strength val- ues, the variability from the moving ^ .0 n=4 2.2 __ r n = 8 n= 16 >- UJ i > F F =1.95 1 i — 1 1 U. ii < i > ' < 1.8 — CO LL -*- o DC - 1 - O 1.4 — h- O < LL Accept F | 1.0 Reject F I FIGURE 8. - Ninety-five percent confidence interval— factor of safety, SW Mine— with varying sample size. averages is representative of the desired reduced variance (fig. 9). A scaling factor, variance reduction, is used to reduce the statistical variance value to the desired variance value. The actual computation of the scaling factor is not as simple as described herein. Involved is a plot of the autocorrection function for the values of concern. Next, a curve is fit to the autocorrelation function, and said function is included in an inte- gration that computes the desired reduced variance value. Two major problems in the application of the probabilistic model were found during a recent field test under a Bureau of Mines contract. One concerns the ef- fects of variance reduction on the prob- ability of failure, and the other is the high cost of obtaining strength data using a penetrometer. The variance reduction value proved to be quite severe, effectively reducing the factor of safety PDF to a single spike. The dilemma is that one has gone through the complex mechanism of probabilistic slope stability and has only obtained a univalued answer, as in the deterministic 12 Q LU O < CO 1 LU T > \- < (D > Z _l LU -1 DC < H 1- CO < 0- CO KEY Sp=Point standard deviation of data S r =rSp=Reduced standard deviation s r §i r=Variance reduction factor S=Mean strength value t V^ FIGURE 9. DISTANCE ALONG EMBANKMENT CENTERLINE Variance reduction applied to spatially oriented shear strength values. approach. Only replicated fieldwork will verify whether or not this is the general case. To accommodate the computations for variance reduction, a sequence of closely spaced strength values must be obtained in all three dimensions of the dam or embankment. Cone penetrometer soundings are required to gather the necessary in- formation. This adds greatly to the cost of obtaining the factor of safety. The obvious question to be answered is wheth- er the benefits gained from a probabilis- tic analysis justify the increased cost. CONCLUSIONS A large sample of direct shear tests was taken at two waste embankment sites. Of the statistical model tested, the nor- mal probability density function best fit these data. This conclusion is site spe- cific and should not be imprudently ex- trapolated. The authors are aware that a wealth of data is now available to the mining industry, since penetrometers of various kinds are now being used more frequently. Thus, detailed investigation of the variation of strength with depth can now be routinely incorporated to make computations more meaningful. In fact, the Bureau has contracted for soil param- eter collecting using cone data ( 3) . The Bishop code with the propagation of error and probability of failure calculations is available from the Bureau for inter- ested users. A suggested first step towards intro- ducing useful probabilistic concepts into the factor of safety computation is to compute a confidence interval about the average factor of safety value , replacing current usage of a conservative determin- istic factor of safety value. It would seem most advantageous not only to know the factor of safety value but to have a sense of the error in that value and an indication of how good or reliable the factor of safety might be. 13 REFERENCES 1. Alonso, E. E. Risk Analysis of Slopes and Its Application to Slopes in Canadian Sensitive Clays. Geotechnique, v. 26, No. 3, 1976, pp. 453-472. 2. Anderson, L. , D. Bowles, R. Can — field, and K. Sharp. Probabilistic Mod- eling of Tailings Embankment Designs. Volume 1. Model Development and Verifi- cation (contract J0295029) . BuMines OFR 16KD-82, 1982, 233 pp.; NTIS PB 83- 122598. 3. Baecher, Marr, & Associates. Cri- tical Parameters for Tailings Embank- ments . Ongoing BuMines contract JO215018; for information contact D. R. Tesarik, Spokane Research Center, Bureau of Mines, Spokane, WA. 4. Bailey, W. A. Stability Analysis by Limiting Equilibrium. CE Thesis, Mass. Inst. Technol. , Cambridge, MA, 1966, 63 pp. 5. Bishop, J., and N. Morgenstern. Stability Coefficients for Earth Slopes. Geotechnique, v. 10, No. 4, 1960, pp. 129-150. 6. Deming, W. E. Statistical Analysis of Data. Dover Publishing Co. , New York, 1964, pp. 37-48. 7. Harr, M. E. Mechanics of Particu- late Media. McGraw-Hill Book Co., Inc., New York, 1977, pp. 427-442. Denver, CO, 1978, pp. 1-16; available for consultation at Spokane Research Center, Bureau of Mines, Spokane, WA. 10. Lambe, T. W. , W. Marr, and F. Sil- va. Safety of a Constructed Facility: Geotechnical Aspects. J. Geotech. Eng. Div., ASCE, v. 107, No. GT3 , Mar. 1981, pp. 339-352. 11. Lee, K. L. , and A. Singh. Report of the Direct Shear Comparative Study. Soil Mechanics Group, Los Angeles Sec- tion, ASCE, Nov. 1968, pp. 1-38. 12. Lumb, P. Probability of Failure in Earthworks. Proc. 2d Southeast Asian Conf. Soil Eng., Singapore, Sept. 1970, pp. 139-148; available from authors of this report. 13. McWilliams, P., and D. Tesarik. Multivariate Analysis Techniques With Application in Mining. BuMines IC 8782, 1978, pp. 22-27. 14. Mittal, H. K. , and N. Morgenstern. Seepage Control in Tailings Dams. Can. Geotech. J., v. 13, 1976, pp. 277-293. 15. Sharp, K. , L. Anderson, D. Bowles, and R. Canfield. A Model for Assessing Slope Reliability. 60th Ann. Meeting, Transportation Research Board, Washing- ton, DC, Jan. 1981, 40 pp.; available from authors, Chem. Eng. Dept. , Utah State University, Logan, UT. 8. Hoel, P. Introduction to Statis- tics. John Wiley and Sons, Inc., New York, 3d ed. , 1962, pp. 139-145. 9. Jubenville, D. Limit Equilibrium Slope Analysis and Computer Software. 16. Vanmarcke, E. Reliability of Earth Slopes. J. Geotech. Eng. Div., ASCE, v. GT11, Nov. 1977, pp. 1247-1263. 14 APPENDIX. —DATA COLLECTION, SLOPE STABILITY, AND LABORATORY TESTING DATA COLLECTION--CDA A detailed investigation of the im- poundment facility was not undertaken, since the scope of this project was to determine strength variability in the beach material and to model an embankment consisting solely of this material. The embankment was constructed by the upstream method using a starter dike of borrow material. The dike is built against a steep side of a valley with an overall downstream slope of 2:1. The im- poundment area covers approximately 7 acres. Tailings were deposited from spigot lines extended from the crest of the embankment. They contain from 15 to 50 pet material passing the No. 200 U.S. Standard Sieve. In general, the coarser material settled out close to the crest, but a layering effect became evident while sampling. Fluctuating pond size and different spigot locations could have caused this phenomenon. for the computer model. These were se- lected to evaluate a potentially unsafe condition (that is, P (failure) > 0) and in no way reflect the actual conditions of the embankment from which the soil was extracted. The location of the phreatic surface, based on the 10 pet freeboard headwater, was determined by the finite- element method (4). 2 No tail water was assumed in the model. The material in the embankment was assumed to be homoge- neous to the extent that no layers were coded into the model. The code was exe- cuted using 74 sets of strength parame- ters (c, . DATA COLLECTION— SW Data were taken along the perimeter of the embankment 15 ft from the crest at intervals of 150 ft. At each data loca- tion, four thin-wall, 3-in-diam Shelby 1 tube samples were taken at the corners of an imaginary square measuring 2 ft on each side. Before inserting the Shelby tubes by hand, the tailings were in- spected for disturbance and cracking. If either of these conditions existed, the cluster of four tubes was displaced slightly to avoid the irregularity or cracks. When the tubes were extracted, the ends were capped and sealed, and the tubes were placed in a foam-lined con- tainer for transportation by truck to the laboratory. Care was taken that minimum sample disturbance occurred. STABILITY ANALYSIS— CDA A slope of 2 to 1 and a phreatic sur- face having 10 pet freeboard were chosen 1 Reference to specific trade names is made for identification only and does not imply endorsement by the Bureau of Mines. As with the CDA embankment, an exten- sive field investigation was not under- taken. A brief, general description will again be presented. The SW tailings embankment has been in- active for years. It was constructed from copper tailings , using the upstream spigoting method. The slurry was about 45 pet tailings by weight; the tailings grind was 40 pet passing the No. 200 Sieve and 75 pet passing the No. 65 Sieve. The embankment is bounded on one side by another tailings embankment. The tailings pond covers approximately 300 acres and averages 300 ft high. The overall slope is approximately 2 to 1. The data for the SW embankment were also collected using thin-wall Shelby tubes. The embankment had from 6 in to several feet of overburden placed on top of the tailings for dust control ^Underlined numbers in parentheses re- fer to items in the list of references preceding the appendix. 15 purposes. The top 3 to 5 ft of tailings was compacted from equipment placing the overburden; in some cases, desiccation was apparent. An attempt was made to in- sert the Shelby tubes using a jack, with resistance provided by a truck. This caused too much sample disturbance so a front-end loader was employed to scrape off the overburden and push the Shelby tubes into the embankment. (The tubes inserted by the jacks were not included in the data analysis.) SAMPLE TESTING Standard direct shear tests were per- formed on a series of samples extracted from the Shelby tubes with a circular trimmer ring. Owing to the time needed to test all the soil samples, some corro- sion took place inside the Shelby tubes. These samples were sawed in half to re- duce sample disturbance when the soil was extracted from the tube into the trimmer ring. The tubes were extracted by hand, sealed, and packed into foam-lined boxes for minimal disturbance. It was neces- sary to lay the tubes in a horizontal po- sition for shipping. The data station pattern is shown in figure A-l. This pattern was chosen so that correlations might be studied based on distance from the crest and spacing distance between data points. Unlike the CDA embankment, the computer model was set up to reflect the geometry and phreatic surface location of the ac- tual embankment. The cross section that was used is shown in figure A-2. The ac- tual embankment had a starter dike that was not included in the analysis. This exclusion could produce lower factors of safety than actually exist. All 130 triplets (c, , y,) were run through the computer code to obtain the 130 values of the factor of safety necessary to calculate the PDF and coefficient of variation. Each sample was consolidated before shearing by applying normal load to allow for drainage. Initial displacement mea- surements from consolidation, along with direct shear test data, are given in ta- bles A-l and A-2. At a minimum, normal stresses of 25 psi, 50 psi, and 100 psi were used. For the CDA data, an addi- tional point of 75 psi was used to deter- mine each , c combination. Owing to the layering effects of each embankment , it was necessary to repeat various normal loads to ensure that enough data were available for a good least squares curve fit for <|> and c. Average shearing rates for the tests were 0.305 mm/sec for the 25-psi normal load, 0.275 mm/sec for 50 psi and 75 psi, and 0.25 mm/sec for 100 psi. The slow rate was used to pre- vent pore pressure buildup and ensure drainage. Cumulative grain-size distribution tests were run for 14 tubes from the CDA data. At least one tube was tested from each data location in an attempt to • • • • • • • • • • • • —I k-25' 324' Embankment! crest FIGURE A-l. - SW data station pattern 16 represent the range of data collected. A range of distributions is shown in figure A-3. COMPUTER PROGRAM FOR PROPAGATION OF ERROR The computer program used to the coefficient of variation of tor of safety is a modified vers program written by Bailey (4). gram uses Bishop's Simplified Slices to determine the minimum safety for a given embankment The failure surface is assumed arc of a circle. calculate the fac- ion of a The pro- Method of factor of geometry, to be an The program divides the slope cross section into vertical slices and evalu- ates moments about the circle center using the force vectors at the base of each slice. The horizontal side forces and vertical shear forces on each slice are not used in the calculations. The number of slices is input by the user but may be altered slightly by the program so that slice boundaries coin- cide with selected points on the embankment . 500 The slice boundaries are determined be- fore any trial circles are analyzed, so the number of slices in each sliding mass varies with the radius of the trial cir- cle. To determine a minimum factor of safety, the program evaluates 1,331 trial circles. The center of each circle is a point on an imaginary square grid placed above the slope of the embankment. At each point on the control grid, the pro- gram computes a minimum radius; that is, a radius that will just touch the slope and a maximum radius based on the geome- try of the slope profile. A series of factors of safety is computed, starting with a radius slightly smaller than the maximum radius. Each successive radius is reduced by approximately one-eleventh of the difference between the minimum and maximum radii. Coefficients for evaluation of equation 4 are required for input in addition to embankment geometry, phreatic surface in- formation, and soil parameters. The pro- gram outputs the coefficient of vari- ation and standard deviation for the min- imum factor of safety, the probability of failure (assuming normal statistics), and 200 300 400 500 600 700 HORIZONTAL DISTANCE, ft FIGURE A-2. - Idealized profile-SW embankment. 800 900 1,000 17 100 90 80 ? 70 o a 60 a» so o S 40 30 20 10 U.S. Standard Sieve sizes 30 40 50 70 80 100 140 200 325 400 \ / / / N \ Y / U ' / // 9, // // t£ \ h J >; & ! / i i // t '/// ti ' V / 5 4 3 198765 4 .01987 6 5 4 GRAIN SIZE, mm FIGURE A-3. - Range of grain-size curves for CDA data. 10 20 30 40 50 60 70 80 90 100 0) (0 « o o o a CO O .001 the contribution by term (equation 4) to the standard deviation of the factor of safety. The pore pressure in the computer pro- gram is calculated at the base of each slice, based on static head. Similarly, the contribution to the error term by pore pressure is on a slice-by-slice basis. Since the user inputs the coef- ficients of variation, it is convenient to calculate both the variance term and the covariance terms for pore pres- sure. In our calculations, it was as- sumed that the correlations between pore pressures and the strength parameters = 0, and that the correlations between pore pressures for any two slices = 1. The covariance terms for the strength parameters are included; that is, r c x r c,Y, and r ,Y. 18 c o ■H i-l CO CO -, d u j_> 0)0 0) H 4J K-l ■H O to HI U li) tn K-l is ps CO ps is CO vO CM #— i 00 -3- o o CM CM CO CD vO -3- ~3- ■3- 00 vo o CD CO CM CD >s 4-1 4J T-l K-4 CU to CJ 3 c a CU TJ IS CD CM in rs —i o vo O — i O -* -I tN CN N CO ~H vD -3- CD —I o o m CM CM -H rs vo is — < CM vD CM CO CO ~H CM — 1 •-< CM O CO O^ G\ cd ^ in O CM — i — I O vO 00 00 CM cm in CM -3- ^ CO -3- -h cm m o in oo co CD CO -3- CO •—1 CD CO ^\ 00 O ~h vO CM CD <3 00 — 1 O — 1 -< HPlOlO CM CD co r^ •3 n tn-j O O O O co m is — i -3- is CO CO o —I —< CM oo in is -3- -h is —I CM CD CM CM —I 00 CM -H O in is m oo is is in co —I —I —I CM CO ^ CO i-l CO cu to cu i-i a CO Cft N N O i—l CD CO CO O m o cm CM CO

-3- co m mo cd CD sD CO CM CD CM CM CO r- o r~ co 00 r^ -3- i-i hio in n CD -3- CM 00 i^ m co in CO 1^ |v vO hco m s O O O rv CM 00 O 00 CM Pv O 3 o CJ •H 4-1 i-l a. C CO 4J H CO in o O rs — i O 00 CM CO O CO vo rs o co cd c\ — I <■ N n rs H(MN CD CO vO N CON rs 00 vO CM -H CD IDO) CO CO CD CD O CM CM CM -H \D vO 00 VD O CO 00 00 CO •* CD CM O mO r^ o CD r*. T^ CD cd oo o in O ~h o m CD r^ in -3- co co n oo o m vo o vo r^ cm >3- m -* o o N \0 N CO P*- CO CO CD m CM VO CO ^\ &\ ~H vO •* in in -3 CD 00 O iH O O o in o CD 00 o — i CD O m cm o CO CM O 00 CD O O vD in CM CM CM 00 vO CM CD CM CO CD Is 00 CD Ss U o 4-1 CO IJ .O « < a o i 1 ■H >v HT3 4J O CU >si-4. K-4 CO 4-1 kl CO U c to t3 c a O TJ HI O TJ h vo vO i£ 2 CJ o o o o co en h -h cm cd in co CM CM CM CM o o o Oi O CM CM CO O CM CM CM O O O \D \0 CM CD -3- -3- HN CM o o o CD — i ~h 00 CM CD — 1 CM —t O O O O — i O 00 vO vO •* CO o CO CM CM CO O O O O vO CD -4 O 00 00 s 4-1 >-, 1-1 K-l U CO O O C O. CU TJ vO CM m CM — i Is CD CO o co m co CD CD CD CD OiOh cd co oo ps co m cd cd cd cd co m r^ cd in co oo m CD CD CD , K-l 4-1 •H 1-4. CJ > CU CO a i-i en bo -o CO CO CO » co co co N CM N CM CO CO CO 00 00 oo CM CM CM CO CO CO oo co oo CM CM CM CO CO CO co oo oo CM CM CM CO CO CO CO oo oo oo oo CM CM CM CM CO CO CO CO oo oo oo oo CM CM CM CM CO CO CO CO CO 00 00 00 CM CM CM CM in in m in oo oo oo oo CM CM CM CM in in in oo oo oo CM CM CM in in in oo oo oo CM CM CM in in in in oo oo oo oo CM CM CM CM ■i cu - H M 4J tO 3 C H 4J CU 4J < -> CO 4J O H 1-1 C CU s o o t -1 CJ -3 O O O » 00 -3- 00 3- — i vD vO M CO CM CM o o o CD CO CM m co o CM CM CM o o o vo in cd cd m vo —I CM CM o o o CM -4 — i O CM CD (M CM -4 o o o o —i oo oo cd *C -3 00 CM CO CM CM CO o o o o vO O -4 -3 00 — i -3 1 r~ CM CO CM CM O O O O O CD -h 00 vO O CM CM CM CM CM CM o o o o m in o is is o cd m — 1 CM -H -* o o o vO-3 CM O CM — i CO CO CO o o o 00 O is in is o CM CM CO o o o o 00 — 1 CD CM O st \0 is CO CM CM CM to o e •H TJ i-l l 4J i-l 4-1 C i-l O CO C > P C I-l MOHCO n oo -3- oo MXIOON co is 3 oo r^ oo CM CD m CM \D ~h CD r^ oo -i co m h CO —I sf CO — 1 CD O O CD CD r^ O -h O -I CD O CD CD CO CO O CO Mnoio CD CD 00 00 VO vD COst o ao 3* NCOMIO O O O O 1 1 1 1 is m o oo <• m m co o o o o 1 1 1 1 <• CD CO vO 00 CD O O O 1 1 1 is —I CM CO is CD o o o 1 1 1 00 vO O — i in rs vo oo o o o O i i i i cu ■-I en i cm co *"* vO 00 en ON UI p^ ao On oo 91 On on O = O = Z CM - Cf> CM O O O O ON ON O ON O 1 ' CM CM 1 ! CM 1 ' CM O X c oo cm a CM CM _ * 00 in en —• P^ en O O _ vO r* On _ -» en nO co 00 X r*- _ » 3- ^n m 00 ON ON NO r** O O N ~ ^4 P- m -~ X •a- — 1 — . NO On ON in m CM n 00 x> inN^o-Na-r-^CT^ co^o-J^^^H H (M IN fv) e^CTvr^in r-« ^ >o f^ n o OOOr^. o^ <*** cn \o o- n r% m on p*> «-i oo vO 00 vo m o> m »— 1 co cn co in o »o -^ vC 00 00 r*. c^ ri n - * in in co 00 vu in in in -j vo ro h n nO ^ -h n 00 ■vf o n -vt n .^in mvocoo ocsicnco n 00 m o oj O 0> OOCOCOOv On ON 00 CO (T> & & en en -a NO DC C i — 00 CM in m CM CM O -a CM p^ -T P^ CM nD CM On no CM CM p~ CM O CN n 00 m O o< en -a NO CM -a :D NO in no CM CM -a CM r^ CM CM O NO CM CM -a m O cn -a c c — x -o .0 — o c o o 00 in *a on sO — 1 O- O — O rn 00 CT* s **i OC 00 00 — « in o*n nnO cm n n in o> 00 00 00 ON B0T3 • • • « • • • • 4-1 1-1 c in > NO r-~ 00 CN ^_, in ON CM NO r-~ CO r-~ CM CO r^ NO ON CM CO •H in cu CO o nO co CN CM r^ i — i co NO CO r>» 00 co ON 00 ■ — i I-l a nO o O CO CM in 00 r~ ON ON CO ON O NO ON NO NO NO ON NO 00 -a- O ON NO in > m co o CM CO NO m CO m in m , r-~ in m o •j- -* CO NO m ^ r~- CO ON ,tI ' CO J-> U CO CJ nO CO O 00 <* in CO in r^ 3 a U * o o o o o o o o O o o o o o o o o o o o o o o o o o O o o o o o O O O O O O 4-1 01 0) 4-1 4J CO r< CN r- oo O o o cd 4-1 —- ( CM CM CM CM CM CM 1 — 1 CM CO CM CM CM CM CM CN CM CM CM CM CN CM CO CM CM CM CM CM CO CO CO CO CO CM CN CM CM CM 53 CJ >-. On r^ CO O ON NO p~- oo ,_, ON •* ON -a- 00 NO ^ m O ON ^D ON _, in On ON CO nO ON n in h on NO co r-. oo on no 4-1 nO .-rt M-l U CO o r^ o CO m -3- OO m o ON 00 in ON * NO ~H in ■* CM o -icOntO r~ r*~ no -3" on co o 4) a. en o ON ON ON ON ON On 00 CO ON On ON ON ON ON ON o ON ON On On 00 ON on ON On ON oo i — oo oo 00 ON ON ON ON CJN XI CD CI) S-i •H 3 •H in o m o m o n o m o in O m o o m o in o o in O in o m in o o m o o m O m o o m o .-1 CO CO CM U0 r^ o CM in r-» o CM in r- o CM in m r^ o CM in in r— O cm in r~~ p«. o o cm in m r~ o cni mms o a CO Cu rt ~H 1— 1 t— 1 -H i-H .— 1 .—1 ~H 9- 01 n NW 4-1 co CO CO CO CO CO CO CO CO co CO CO CO co CO CO CO CM CM CM CM CM CM CM CN CM CM CM CM CM CM CM CM CM CM CM CM CM ■H O o> •H > cd oo 00 oo 00 00 00 00 oo oo 00 oo oo oo 00 00 00 oo 00 00 CO 00 00 00 00 00 CO CO 00 00 00 00 00 oo oo oo oo oo oo CN CM CN CN CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM o, i-i w bO a) ^ H u 4-1 cd ■s a o o o O o o O o O o O O o o o O o o o O o O o o O O O o o o o o O o o o o o i-i 4-1 01 4-1 00 r* 1-^ CO 00 CO m -* o o r^ r^ -tf NO ■H 4-1 oo f~ r» o 00 r~~ CO ON ON ON oo r~ 00 ON r^- 00 oo r~. 00 f~ oo 00 O ON CO CO 00 ON -h CM O —l O CO CO CO N CO ■H o cd > u o CM ^ H ^H 1 •H CO in O oo o CM ^ -d- 00 CO CM 00 t~» ON CO m -d- CO _ Cf> p^ in cm CO O in _ m _ _ co c ps. in o CM O C| CM _, Ps O CM _, 0\ on o CM nO O OO CO m Ol CTN ^.o p*» BN S3 mo p ^ cc ■ st UO '"*■ CM sT MO CO CM s3 c 00 CM ^-r CO m ps CM X CO _, N sT in -3- - .- on O o •» M0 oo CO U"l CO CO 00 oo mo X ■Z CO o> ao a U0 co uo MS S3 in in m ps CO ifi oo in P~ co v£> M3 CO r^ r^ m m CO o> CO CO -3 On CO UO o> c~ r^ r^«. *n u-\ in o> o*\ o cr* o> CMJnO o C7\ O^ C^ O^ CT^ c a C - o O O a o O C a O O O CO o O o c O O o O o o o o O o o O o o o O o O O o o o o O o M0 -~ t— o> ac in -3 - * CO ON -~ CO co m CO -3 — r~ C-! ON CM -^ ■a r-~ lO — CM r^ 00 m oo CM CM CO ■3- CO in CM MC CO o -H P~ — ' UO -3- K CO 0*v o ~^ r^i sT M 01 • • • • • • • • • 4J 1-1 C T3 in r^ vr sO 00 |N sO sO IN c u CO CO CO CO CO CO co CO CO CO H M-l (0 iw 3 O i-l >s CO oo 00 IN 00 oo sO in c M 4-1 in 00 CO I — 1 t— 1 CO CO .—i rN Q) o 01 CM CM o OS 00 in CM o r~ .-i 4-> CH ^H CJ CO . — I CM .—1 i— I CN • — i s oo OS rN in r-^ CM so cni o CO 00 -a- CM CO 00 o ^ ^ sO m IN r» m 00 IN m 00 o CO CM 00 ON -a- 00 a 00 CO m OS , rN CN -J" so CO Os o m CNl ^ oo m in s iH CU H CO CJ sO n- -a- SO in CM CM CM OS ON -a- in -a- IN o o CO sO o -a- sO CO UN CO -3- CO OS CM -3- o m in <■ ON in sO IN IN o c

os ON CO 00 ON os Os 00 OS ON Ol 00 ON ON ON ON OS OS ON 00 Os 00 ON OS OS OS ON ON ON 00 ON ON ON ON -o 0) 0) M 1-1 s •H UO o o in o O m o in o in o in o m O in o m o m o in o in o m o m o m o m m in o in o in m o ^4 CO CO CN m m r^ o O CM m r-» o CM m IN o CM in IN o CM m IN o CM m IN o CM UO IN o CM m r- IN IN o CM in IN IN o D. en a. ~H a. 0) s <4-l 4-1 0) CO CM CM CM CM CNl CM CM CNl CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM CM a. U CO 60 a> ^ rH u 4J to 3 c o O O O o O O o O O o o O o o O o o O O o o o o o o o o o o o o o o o o o o o o o •H 4J 01 4-1 CM M ^ ~* ^ " ^H 1 e 00 in ON CO 00 CM ""< CM 00 O CO ON oo 00 in O st r- o O ON co CO CM m nC o in NO co -3 CM _ On U-N co p** in X IT CM -3 _l _ o CM m r~ XT NO on r~- ON go m \o 00 co CM m _, O co ON o o -3 ON CM o CO NO CO 00 co O a ON o CM CM. CM CM CM O CM CM CM CM CM CM CM CM ■3 go ON On rsi -3 O ao ON r^ <- CO r*. p^ NO o «3 3 ie vo P"* ON CM CO in r^ CM CO MD 00 CM CO in CO r-~ ^ H CM SI r^ r~ oo ■"* CO in no _ ON ON ~3 -3 p^ CM p*. l — 00 X go X , O O o t CO o NO ^r O O ON r> CO C X ON o NO CM DO r** -3 1 — — X cri NO oo NO o iC 1 — O st CM o o o 1-1 . ^ C o o o o o o O n fO O (N ^ — m CM —• a 00 CM CO co in CM CO -3 CO CM CO CM NO CM CO CO CM B" CM 00 CM m ~3 CM CM CJ CM 00 CM m CM NO CM ON CM CM P^ CM m CM -3 CM CM ON CM NO CM r-. CM O -3 CO CM CM CM CO CM CM CO CM CO co O CO CO ^^ CO !■*» in CO o> ao CM m X X X NO CO 00 m CM CM in o c 0" O in m co NO in X' nO CM CO ON CO CO O NO 00 CM CM CM NO O CO in o CM X) X CM co NO ON CM CO ON CO o to 00 00 P^ NO NO -3 p^ X> CJ ON CM OC 00 ^ oo \r\ o it* O eg m r^ O u. iri o O ^ o N N ia in fN. o in iAO m O in N in n o in o ia in o o (N lAN N OO m o m o cm in r- o c~ co -- CO CO co CO CO CO CN ~3 ~3 n9J -3 sT T in ON -3 -3 m m ~3 •3 _, CM c 1 _ CM "" " CO •* CO CM CM CO CM 'J CM CM CO CO CM CM CM CM CM CM CO CM •"* CM CM CM cl CM CO CM CO CM CO CM CM CM -3 •* to -.f .- X. ■£> m M3 on p* ON -3 in ^ O CO P~ -3- IT -3 h* en CO 00 O o -3 in CO x. CM -3 a\ CM r^ P-~ CM —i St — 1 ON nO CM -r r s PN* X KM .' nO -3 K *M nO 00 r*. O co r^ CM — - ^H 00 m CO ON CO ON 00 00 nO o H — 1 CO CM o CI P~ NO CM NO at r~ o f^- ON ro St ^T o o o> o< ~ BC B 00 X p-. ao ON ■y ON ON oo a X. ON X p^ 00 00 r~ On 00 O en ON 00 00 O 00 oo on r^ CO p~ oo P^ ON o CI. o ■"• — — — ■ — ■ "- , ■"■ -• c » -3 -3 >» st -3 CM CM CM CM CM CM ON O — < CM st in in m CM CM CM CM 24 0) 3 d c o o I >N 14 o 4-1 « o .o c o 1-1 1-1 CD CD CD P. O -3 o NO ON O o ON oo NO NO CM oo -3 -3 -3 fv -3 NO in CO Internal friction angle , deg in in NO CO 00 m -3 CO ON NO rv CO ON rv ON CO m CO o -3 ON r- oo CO in -3 -3 in in CO Fellenius factor of safety > 1.775 i 1.028 > 2.159 r- CO in *—* > 1.642 > 1.736 > 1.395 > 1.006 Wet density, pcf CM vO ON O O O «o -3 -3 -3 co -3 oicocn ho o HrfN NHCM 118.89 111.92 121.95 118.50 122.95 123.18 116.91 119.78 99.72 103.75 110.84 104.83 -3 O O CM CO -3 •— i *— i ^H f— 1 111.80 105.53 105.73 115.56 112.62 113.51 118.58 119.86 120.32 95.78 88.96 111.76 114.20 120.94 111.80 106.50 110.64 100.07 108.16 Peak stress, psi rv cm cm r- co on rv in cm in co ~h o -3 cm co m on co cm -3 no on rv 18.27 35.97 52.11 70.34 rv co — < in *-" -3 CM ON CON MO CM -3 NO 00 24.00 43.78 52.11 64.61 O tv ON -3 — 1 NO r- oo oo rv on oo o -3 o no NO 00 -3 ^ CM -3 NO ON cm on so in in m co -a- o on -3 no co co in CM -3- |v vO 00 O rv *-t co -3- O 00 CM — I no -3 in rv co o CM -3 nO O ON t—i ON CM 00 -3 O r— o cm o on CON vO CO h -h co in con Initial satura- tion, pet m rv o o o oo O ON O O ON NO co cm o o oo rv 00 00 O O 00 00 •— » .—i 91.90 100.00 100.00 91.28 100.00 100.00 88.67 83.10 vO ^h O CM O vO r-*. r^ o u"i O r->* ON ~* O O O vt P*ivO O lAO vO O ON CO CM ON CO -J rf oo on m m m -3 -3 NO OO 00 -3 ON no ON •— i 00 CO On rv o no cm oo ^ON IN DON no oo -3 r> m ON 00 O ON ON — i o cm <■ rv co cm no no rv — i no no rv o no on rv on r-- oo o no m on i — no no co in Consoli- dated dry density, o r- •* co-* on OONNvOCOH CM IV CO CM r-< CO O ON O O O O •—i .-H t— J r— ) —4 98.81 103.21 98.25 101.01 101.73 103.02 100.10 101.83 92.36 91.58 93.51 96.83 CM —I NO CM O CM O O •— 1 F-H ~h oo in no no r~ -3 on 00 NO ON -3 ON ON ON O ONl/N in H N IN H VO CO 1 — . |v ON -3 in -3 ON ON o o o p- o m o -3 NO O ON 00 NO cm co on h in On ON ON o o N- CM CO ON CM CM CO ON CM -3 CM CM I-. NO NO ON ON ON ON ON Moisture content after test, 4J CJ o o o o o o rv no oo on o oo .— i in in -a- no co CM CM CM CM CM CM o o o o in rv on m O -3" -H o CO CO CO CO o o o o NO CO CM no -3 -3- CO -3 CM CM CM CM o o o o o o m in co co oo in -3 -3 00 1 — CM ON hniOhio h o o o o — i O 00 ON 00 no -3 00 o o o o o ~h in in cm co HO N-IN CM CM CM CM CM o o o o o — 1 CM NO NO -3 CM 00 ON ON — < — 1 —1 -H CM o o o o o ON CM -3 nO in 00 CM CM CM O CM CM CM H CM Dry density, pcf oiNNOom CM CO — 1 CM — I CM CO •* 00 N CO N ON ON ON ON ON ON 91.10 83.09 94.45 90.81 00 O O CO NO *H ON «— 1 00 ON -3 nO ON ON 91 ON O co -3 rv ^H CO -^ CO MOOON 00 00 OO 00 CM tv — i m m in 00 ON rv oo O on VO ON HH -3 O CM P- ON ON ON ON ON O O ON 00 ON CM 00 00 CO CM -3 NO 00 CO ON ON ON ON ON -3- CM m ON CM -3 CM -3 -3 no in cm co in on 00 00 ON ON ON -3 in on r-- no r- — i co oo r- nO r-. o oo on 00 00 ON 00 00 Applied pressure, psi in in o m o o cm cm m rv o o i— i f— i in o in o cm in tv o m o m o cm m r- o m o in m Nm m — o o o o ~H i— 1 m o m o cm in rv o in o in in o cm in rv rv o m o m o o cm in iv o o in o m o o cm m i-. o o o 1-1 >N ■4-1 4-> 1-1 iH o > V CO a, u C/3 00 CO CO CO CO CO CO oo oo oo oo oo oo CM CM CM CM CM CM CO CO CO CO oo oo oo oo CM CM CM CM CO CO CO CO 00 00 00 00 CM CM CM CM -3 -3- -3 -3 00 00 00 00 CM CM CM CM -3 -3 00 00 CM CM -3-3-3-3 00 00 00 00 CM CM CM CM -3-3-3-3-3 oo oo oo oo oo CM CM CM CM CM -3-3-3-3-3 00 00 00 00 00 CM CM CM CM CM -3-3-3-3-3 00 00 00 00 00 CM CM CM CM CM Initial moisture content, pet O O O O O O -3 no in oo cm co O O O O in o co in O O O O rv 00 O no O O O O in m cm m O O — i m O O O O — i m oo on O O O O O i-~- oo m o co o o o o o ^H in nO no -3 o o o o o on cm no no in co in on — • oo in CM CM CM CO CM CM O CM -* o CO -3" CO CO oo o rv -3 CM CO CM CM -3 -3 -3 rv HN-J -H — 1 ON -3 —l 00 NO -3 00 -h ^H CM CO CM CM CM CM CM CM CM 00 ON ON -H —i — l — I CM 00 CM CM CM O CM CM CM -* CM Initial void ratio rv co o oo oo rv JN rv o ~H ON *-< rv oo oo oo oo oo ON NO ~H \0 CO CM —I -* ON t— 1 ON ON O CO CM 00 ON 00 NO CO rv r-. oo oo NO 00 CM -3 CON -n 00 OHCMOV co m oo m o 00 co on m -3 P- -3 CM CM oo on on oo NO CM CM CO CM O 00 CO ON o ON 00 00 in 00 in no rv i — o r-. in on in oo oi cocoin -3 -3 -h in in -3 co no on r-. O O ON ON ON o •— 1 t— l »— i i— i •— i ~H ~H ~* f— 1 Initial consoli- dation, in (i» 1 — Ch CO H N ■o co -3 in oo m o o o o o o oo in on — i rv on m o o ■— l o •— i O 00 CM NO co co in m o o o o n o co rv in on -3 rv o o ^ o ■3 m in no -i O O O -3 -3 -3 no rv rv o o o o OlOlOHH ON r>. oo CM ci O "* O -> r^ O vC 90 CSI -3 vO in X r*. ON ON in NO ON CM CM r*. P^- in r^> r~ ON CM o CM CM CO ^ —, _< _i © ON CO 00 _ H — ■ ^i _ o 3 o CM CM CM CM CM CM CM O) CM CM CM CM CM CM CM CM lO 90 ON p~ o DO o o NO X o M3 r^ «* nD ON ~o -3 -3 x> NO CM ON nO CM -3 ■3- •3- O o NO 00 -3 CM NO r» CT* in 0* CN — CM in X ON ~3 NO CM m ON X X o r- — < CM r^ an "-< —> CO — ' ON CM CM n3 1-1 -3 -* r~ ON -3- CM ON CM rsi 90 a -^ on on vO 00 — __ a NO -3 CM in i — NO 00 © in ON on _, ro ,-^ ON CM CM m o on _ ON m CO CM CM o _H -. ro >c vO '"* ™ co in r-. CM 3 00 0> o o on o *"* nO X 00 o o ■X ON on ON ON 00 NO ON ON oo 00 ON ON O C_N o 00 CTN o on ON ON O O DO -3 ro o ro o •* in r-v CO CO r- m CO NO r~ lO _ ON ON 00 rv CM ON «3 00 in MD 3 3 O ON © © © ON On C on ON O © o ON O o o ON o c © r^ O O O o o o o o ^ O O o O CM O O O o O o o © O o © © o o o o o o o © o o o O O o O © o Q o o o o O O O o o o O in -. 00 f*N 3 I**. o -3 3 ON ON CM ON NO r^ CM ON lO ON ON r— m ^ CN CO NO , O <3 r- r^ CO ON on . o NO ON x> O ON ON ON X ON ON Oi ON ON o ON ON ON ON ON ON ON ON ON ON On ON ON .- o in o O in o in o in o m © m o in o m © u-, o in © m m o m O m m o in o m o in o m o -. m r^ = c CM in r* O CM 1/1 r-- o CM m r-- o CM m r^ o CM in r* o CM m r^ r-. o CM m r*. o cm m r^ o -7 - -3 - 3 CO o © —i fNN- rNi CM -3 C i O r-~ nO o X) ao — i -3 on O o O CO -- »H c^ c CH p* O ON O ON on ON O C ON — ao DO ON ON ON 90 ON an ON oo X r^. r^ on p*l 00 X an on en on an oo on Xj CO w * — — — ' -* •" - CM .- - in a ' I r- r-» c o © m nO CO CM -3 'J -3 o o m e -£> • ON m 00 in CM r- O CM e 00 r - -3 - <3 a f : - 1 LT nO j~ m— -3 CO T r^ CM r^ 90 CM © c r^ 9 r^. ON vO oo — 1 o —t a x> ^H ~* no an o — i □ a i i ' i 1 o 1 a i ' ' O a © 1 1 © c ' ' o 1 O l o 1 o 1 o 1 o i O I 1 ' i O i O i ' i o o ■ i i i — - .' o r^ DO i © 'j c -3 m e h* X ON o •vi ro I m sO rs 00 a o CM ro -3 m nO r~ 00 o- a 9 ~ - o» T ON 0> ON © a a o o o o o © o CM CM CM CM CM Cm CM CM CM CM en EN CM CM ' 1 ' i CO ro — " CO ro CO " m CO "N CO CO CO CO m ro ro ro CO en ro CO CO CO ro CO ro ro ro en 26 TABLE A-2. - SW Laboratory data Fellenius Wet Internal Cohesion, Fellenius Wet Internal Cohesion, Sample factor of density, friction psi Sample factor of density, friction psi safety pcf angle, deg safety pcf angle, deg 1 1.948 111.36 39.39 4.22 66 2.107 100.08 43.20 2.60 2 1.932 103.54 39.44 3.96 67 1.875 114.14 38.10 2.30 3 2.124 103.67 38.51 9.16 68 1.674 105.11 39.50 .00 4 1.899 115.81 37.13 2.40 69 1.904 108.95 36.31 7.34 5 1.838 112.50 36.25 2.40 70 1.775 111.33 40.98 .05 6 2.008 105.31 36.80 5.52 71 1.871 108.26 39.54 2.40 7 1.969 95.74 36.42 5.26 72 1.801 122.41 36.60 2.60 8 2.137 100.14 44.04 .57 73 1.790 119.02 38.60 1.20 9 2.017 103.96 37.51 4.74 74 1.964 107.91 41.00 2.40 10 2.245 99.51 41.65 3.70 75 1.819 107.18 34.80 7.30 11 2.402 102.52 42.03 6.30 76 1.913 110.10 39.80 1.60 12 1.937 111.54 43.19 .00 77 1.913 105.47 40.40 1.20 13 2.009 96.73 36.20 6.30 78 2.040 112.15 41.90 2.60 14 1.713 105.49 31.90 5.50 79 1.887 110.95 40.80 .60 15 1.550 104.15 36.91 .00 80 1.661 114.64 35.20 3.30 16 2.246 104.24 43.23 1.61 81 1.622 109.11 34.90 2.90 17 2.388 106.46 42.78 4.74 82 2.363 119.12 43.80 6.90 18 1.820 103.97 34.56 4.48 83 2.288 112.54 47.00 .50 19 2.328 109.54 37.78 11.25 84 2.030 114.79 45.00 .00 20 2.609 103.14 43.77 13.59 85 2.102 106.93 42.40 3.40 21 1.909 110.85 39.89 2.66 86 1.916 103.65 40.14 1.35 22 2.055 104.04 42.71 2.05 87 1.984 104.89 40.24 3.96 23 2.169 97.81 44.12 2.66 88 1.962 102.07 39.95 3.96 24 2.216 103.96 40.59 8.38 89 2.524 113.11 43.23 11.51 25 2.134 100.61 42.96 1.61 90 2.144 109.08 37.78 10.73 26 2.075 102.06 42.82 2.40 91 2.096 112.27 36.69 11.25 27 1.900 108.50 43.10 .00 92 1.771 98.86 41.10 .00 28 1.896 99.67 36.28 7.99 93 1.980 100.70 41.90 1.80 29 1.812 96.12 39.59 1.61 94 2.041 100.83 44.10 .30 30 1.965 105.12 41.27 2.14 95 1.802 105.89 41.60 .00 31 2.617 111.45 42.90 15.50 96 1.987 107.55 40.80 3.20 32 2.194 112.55 44.50 .90 97 2.126 111.88 35.60 14.80 33 1.802 104.27 34.06 7.73 98 1.940 111.79 43.70 .00 34 1.998 115.73 37.82 7.86 99 1.928 100.15 40.90 2.10 35 1.981 108.40 35.98 9.42 100 1.974 111.59 37.50 7.40 36 1.773 103.94 41.13 .00 101 2.212 102.85 43.86 4.70 37 2.044 118.10 33.80 15.67 102 2.204 97.90 42.91 5.00 38 2.150 112.20 42.27 2.40 103 2.110 99.18 41.89 4.48 39 1.740 109.56 33.15 7.60 104 1.894 105.49 37.56 5.52 40 1.716 114.06 40.20 .00 105 1.945 98.80 41.17 2.14 41 2.360 104.86 40.60 12.80 106 1.836 111.97 27.10 18.79 42 1.602 109.29 31.00 7.00 107 1.923 102.53 31.00 15.30 43 1.971 115.27 41.30 1.30 108 1.574 111.19 31.80 5.50 44 2.028 101.66 39.80 5.50 109 1.960 101.19 36.10 9.70 45 1.674 113.76 39.50 .00 110 2.098 104.65 39.70 7.10 46 2.441 108.66 45.30 6.20 111 1.742 111.58 35.80 4.40 47 2.033 106.11 41.10 1.90 112 1.547 107.24 37.30 .00 48 2.228 106.41 40.50 8.80 113 1.508 106.87 36.60 .00 49 2.199 114.94 42.20 5.70 114 1.340 107.74 29.50 2.90 50 1.809 114.08 41.70 .00 115 1.793 98.32 38.20 2.80 51 1.771 96.29 41.10 .00 116 1.517 102.58 26.80 10.50 52 1.975 102.36 36.80 8.12 117 2.137 99.20 40.90 6.30 53 2.170 98.83 41.22 6.56 118 1.570 99.17 30.80 7.20 54 2.041 100.47 41.99 2.92 119 1.798 111.45 37.30 3.70 55 1.460 111.94 31.40 3.40 120 1.938 112.07 40.50 2.40 56 1.868 109.85 36.60 6.20 121 1.983 105.09 41.00 2.90 57 1.816 108.53 38.70 1.00 122 1.633 112.18 25.90 15.00 58 1.884 109.03 39.80 2.30 123 1.473 114.27 28.70 6.80 59 1.847 104.07 42.30 .00 124 2.057 112.02 41.40 4.20 60 1.826 107.41 39.20 1.10 125 2.063 108.98 38.20 8.40 61 1.603 106.65 38.30 .00 126 2.418 107.78 46.60 1.90 62 1.816 109.90 38.30 2.80 127 2.181 106.15 44.40 1.20 63 1.679 109.93 39.60 .00 128 1.476 110.83 33.00 1.90 64 2.028 102.86 41.10 1.80 129 2.154 100.33 46.70 .00 65 2.013 99.50 40.60 4.20 130 1.740 94.96 40.60 .00 INT.-BU.OF MIN ES,PGH.,P A. 26Q9E A" • v 6 ^ » ^ s* > 4<2* O. ' . „ o ^ .: ^6^ V c*J^w«' °^ 4P .S^r Q_ * .■» o >^ *1*°* k V^u *bV" ^ -v "of ^^ .^ ♦: \f -o *r.T*-A ^ *-"^T ,, .G*' ^t k "*•'•» ,, "i0 , •o V 'J * ;, *w A°+ .1 »p*\. l '- ^^ ^^&: W •JS'' ^/ -Wife w ; Jsfe^ ***** ^^ v* 3 *w ^ v & \0 «5°^ - 1> 6«_-«* *S?^V %^'\S %' 7 ^ ; '^ \ ,; ^Y w • ^o^ . »°v '. ^°* . &* ^-."'••••\0' <*"*^7V^*.G* ^."'••»* 4 ^ V <». **5tv^ , .6 - *' ^*.*' , "»* 4 ^ ^**6* 3". ^-6*