Liquefaction risk analysis at S.G. La Rena, Catania, (Italy) M. Maugeri,^ G Vannucchi ^ Facolta di Ingegneria, Universita dicatania, via Andrea Doha 6, Email: mmaugeri@isfa. ing. unict. it ^ Dipartimento di Ingegneria Civile, Universita di Firenze, via di Santa Marta 3, 50139 Firenze, Italy Email: giovan@dicea.uni.it Abstract The liquefaction risk analysis at Catania city (Cascone et al/), shows that the risk is more relevant at S.G. La Rena site. Three dferent methods for evaluating liquefaction risk, based on CPT are employed. The methods considered are those proposed by Robertson and Wride (Fear)^, Suzuki et al/, and Olsent At present, only 11 mechanical cone penetration tests at the S.G. La Rena site are available. The aforesaid three methods have been applied to these tests and the results were compared. Dferent seismic scenarios with magnitudes of 6.5, 7 and 7.3 on the Richter scale were considered; dferent maximum accelerations, 0.30, 0.35 and 0.40g, were also applied. For an analysis of the evaluation liquefactionrisk,the synthetic index introduced by Iwasaki et al.*, which is representative of the liquefaction potential in correspondence with every vertical explored by means of CPT, was employed. 1. Introduction This study is part of a larger research programme funded by the National Research Council - National Group for the Defence against Earthquakes (CNR - GNDT), The Catania Project, devoted to evaluating the seismic risk of a highlyurbanised area, such as that of Catania which is located in a seismically-active region. The reference earthquake simulates the catastrophic events of January 9 and 11, 1693 (I = X-XIMCS, M~7), which caused as many as 54,000 deaths in
302 Earthquake Resistant Engineering Structures A A ^ A A 0 A O 5PT A CRT A A ^ /x A Figure 2: Plan view of boreholes and of the eleven CPT at the La Rena site. Figure 1: CPT tests location Eastern Sicily. The severity of the seismic design action, the vulnerability of the built-up area, the local geotechnical conditions of some sites of Catania, and the existence of historical documents that describe effects being explainable as liquefaction phenomena, have necessitated carrying out a liquefaction risk analysis. At present, specic geotechnical in situ and laboratory investigations are not being carried out; only geotechnical data existing from the past surveys have been collected and analysed. In the city of Catania, eleven zones have been identied which, on the basis of their stratigraphical and geotechnical conditions, can be potentially liquefiable (Cascone et al/). Among them the risk is more relevant in the zone n. 11, which corresponds to the S.G. La Rena site. 2. The San Giuseppe La Rena site The San Giuseppe La Rena site is located near the sea and near the Catania harbour (Fig.l). It is formed by alluvional deposits underlined by clay formations. Site investigation including 8 boreholes and 11 CPT tests are available. Soil profiles given by boreholes up to a depth ranging from 10 (borehole no.3) to 30 m (borehole no.4 and 6) show a superficial layer offillof thickness 1.8 TO 2.6 m, then a loose fine sand until the depth of 6.15 to 13 m and then a medium dense sand up to the end of boreholes equal to 30 m. Inside the boreholes SPT tests were performed; the results obtained are reported by Cascone et ala The water level is practically coincident with the sea level, which is some meters below the surface. Figure 2 shows the planimetrical arrangement of the boreholes and of the eleven CPT at the La Rena site. Figure 3 shows the average profile of the cone resistance q< of the tests and the profiles of minimum and maximum values
Earthquake Resistant Engineering Structures 303 versus depth. The lines are rather close to each other; therefore the deposit is homogeneous in a horizontal direction, and every CPT can be considered to be representative of the geotechnical conditions of the whole area. 3. Liquefaction risk analysis from CPT At present, the CPT can be considered the most important in situ test for evaluating liquefaction potential. Because it is more accurate and more repeatable than SPT, it is frequently carried out; it is not expensive and it provides continuous resistance profiles. The first methods of analysis, which were developed in the eighties, utilised the procedures created for SPT, prior to converting the cone resistance qc into the equivalent index N<%). These methods required a knowledge of the grain size characteristics of the soil, because the presence of a not negligible fine content (FC > 5%) considerably influences the cyclic resistance ratio CRR of a sandy soil. That condition annulled to a large extent the advantages of CPT, because it required extracting a large number of soil samples for grain size tests or utilising empirical correlations which were not very reliable. In recent years, the growth of the statistical basis of CPT data have made it feasible to develop methods for liquefaction risk analysis directly from CPT results, which are very reliable. Some of these procedures not only do not need a previous knowledge of the size grain composition, but they also consider the other factors deriving from the presence of a fine fraction (soil plasticity, stress history, soil texture). The most credited among these procedures at present is the Robertson and Wride (R&W) method in its most up-to-date version (Youd^). This method has been assumed as reference in this paper In order to evaluate the estimated error due to the method of analysis, the results have been compared with those obtained from two other methods for estimating 100 150 200 250 300 the liquefaction risk from CPT: the Suzuki et al/ method and the Olsen^ method. After defining the profiles of the cyclic resistance ratio CRR, of the cyclic stressratiocsr and of the liquefaction safety factor LSF, the liquefaction potential index PL was calculated for every CPT, so that a direct comparison in terms of hazard between dferent sites could be possible and a liquefaction risk map with equipotential contour lines could be drawn. The liquefaction potential index PL is defined as below: Figure 3: Average profile of the cone resistance q^ and profiles of minimum and maximum values
304 Earthquake Resistant Engineering Structures where z is the depth of the layer and F(z) is a function of the liquefaction safety factor (LSF) which can be replaced by the following equations: F(z) = 1 - LSF for LSF <= 1 (2) F(z) = 0 for LSF >= 1 (3) and w(z)= 10-0.5 z (4) The PL has been categorised in 3 classes, as follows: 0 < PL <=5 belong to areas with low risk 5 < PL <= 15 belong to areas with high risk PL > 15 belong to areas with very high risk (1) 4. Analysis of the CPT Tests Results The penetrometric data of the eleven CPT were processed by using three dferent methods, by way of example and in consideration of the transversal homogeneity of the deposit, only the results from R&W method and limited to test n.l are shown. Figure 4 shows the profiles of the cone resistance qc, of the sleevefrictionstress f, and of thefrictionratio FR of the CPT n. 1. In Figure 5 the data points of the test are represented on Robertson's^ classication chart. The logarithmic diagram has in abscissa the dimensionless normalised friction ratio F: xloo (5) 0 100 200 300-10 Figure 4: Profiles of the cone resistance qc, of the sleeve friction stress f, and of the friction ratio FR of the CPT n. 1
Earthquake Resistant Engineering Structures 305 and, in ordinate, the dimensionless normalised cone resistance: Qc - cr,, (6) The chart is divided into nine zones, which correspond to nine classes of soil: from organic soils - peats (class 2) to gravelly sand to dense sand (class 7), from sensitive, fine grained soils (class 1) to the very stf, fine grained soils (class 9). The curves delimiting zones n. 2 to a 7 can be approximated, in the logarithmic plane, by arcs of concentric circles. The curves delimiting zones n.2 to n.7 can be approximated, in the logarithmic plane, by arcs of concentric circles. The radius of the circles is the soil behaviour type index I* and the co-ordinates of the common centre are: XQ = -1.22 and yo = 3.47. Consequently, to identy the soils of classes n. 2 to n. 7 (classes n. 1, 8 and 9 are excluded), it is sufficient to calculate the soil behaviour type index Ic and to very the appropriate class. Most of the data points of CPT n.l are included in zones n. 6 and n. 5, that is, a soil which is liquefiable with respect to the grain size composition. In Figure 6 the profile of the soil behaviour type index Ic of test a 1 and also the stratigraphical interpretation are shown. Excluding the surface layer, which is about 1.5 m thick and is composed of over-consolidated clayey soil, silty sands and sandy are present everywhere, except from 7 m to 9.5 m in depth, where a layer of silty clay with thin layers of silty sand is present. In substance, the graph of Figure 6 confirms the graph of Figure 5 (obvious), and species the depth and the thickness of the soil layers. Figure 7 shows the profile of the apparent fine content FC estimated from the penetrometric data as follows: Ic < 1.26 1.26 <=!<,<= 3.5 I;>3.5 FC (%) = 0 FC(%)= 1.75 FC (%) = 100 (7) 1. Sensitive, fine grained 2. Organic soils - peats 3. Clays - silty clay to clay 4. Silt mixtures - clayey silt to silty clay 6. Sands - clean sand to silty sand 7. Gravelly sand to dense sand 8. Very stf sand to clayey sand 9. Very stf, fine grained Figure 5: Normalized CPT soil behaviour type chart
306 Earthquake Resistant Engineering Structures The FC explains not only the actual fine content of the soil, but also the effect of the plasticity of the fine fraction. The FC value of the sandy soils of CPT n.l is, in general, in the range between 10% and 20%. In Figure 8, the corrected and normalised cone resistance profile q^n and the corresponding profile of the equivalent resistance for clean sands (qcm)cs are compared The qdn is calculated as follows: (8) with: The value of the exponent n is a function of the grain size characteristics of the soil; it is in the range from n=0.5 for the sands and n=l for the clays. The resistance (qdn)cs is calculated as follows: in which: (10) (9) Ic<=1.64 1.64<Ic<2.6 Ic >= 2.6 Kc = -0.403-Ic* + 5.581-Ic* - 21.63-Ic* + 33.75-1. -17.88 the soil is fine grained (silts or clays) and therefore there is no problem of liquefaction. 0 20 40 6D Figure 6: Profile of the soil behaviour type index ^ Figure 7: Profile of the apparent fine content FC
Earthquake Resistant Engineering Structures 307 The distance between the two profiles represents the increase in resistance due to the presence of the fine fraction. The missing data points of the (qdn)cs profile correspond to the observations of the fine grain sized soils (which in practice belong to classes n. 3 and n. 4), for which there is no risk of liquefaction in any case. The (QCIN)CS profile is used to calculate the cyclic resistance ratio CRR. In Figure 9, the profile of the cyclic stress ratio (CSR) and the profile of the cyclic resistance ratio (CRR) for CPT n. 1 are shown. Only the signicant data points of the CRR profile are represented; the data points of the layers in which liquefaction cannot occur either because of the soil class or of the greater cone resistance, are excluded. The cyclic stress ratio CSR is calculated from the following equation: MSF I g (ID in which: ra=l-0.00765 z %= 1.174-0.0267 z and the magnitude scale factor MSF for z< 9.15m for 9.15 <z<23m is calculated as follows: M<7.5 102.24 / *, f \ 3.3 MSF = 0.5 (12) M = 7.5 M>7.5 MSF=1 MSF- 10* (13) (14) The cyclic resistance ratio (CRR) is calculated from the following equation: (qcinl<50 CRR = 0.833 1000 + 0.05 (15) <160 CRR = 93-1000 +0.08 (16) In the case of Figure 9, the following seismic design data are employed: amax = 0.4g and M = 7.3. The effect of the values of the seismic design data on the risk analysis results will be discussed in the next paragraph. In Figure 9, only the signicant data points of the CSR profile are represented. The resistance demand D=CSR is larger than the resistance capacity C=CRR for only a small but signicant number of penetrometric data at a depth of between 5.4 m and 10.4 m from ground level. Figure 10 provides a graph of the liquefaction potential index
308 Earthquake Resistant Engineering Structures 0 100 200 300 400 500 0 0,2 0,4 0,6 qcln (qcln)cs Figure 8: Profile of the corrected and normalized cone resistance Figure 9: Profiles of the cyclic resistance pressure and of the cyclic stress ratio PL for CPT a 1, which attains the maximum value 6.73 at a depth of 10.4 meters. Therefore, for the design seismic action, the liquefactionriskmust be considered to be high, because the threshold of PL = 5 is exceeded. However, the range of PL is between 0 and 100, and there is no intermediateriskclass between the low risk class (PL < 5) and the highriskclass (5 < PL < 15). 5. Effect of the seismic design data on the evaluation of the liquefaction potential The seismic risk analysis of the city of Catania was carried out in a deterministic way with reference to a catastrophic, historical earthquake, well localised, with estimated magnitude M=7+, that occurred in 1693. Another second-level scenario for seismic events of less severity, but with a shorter return period, was considered. For this, an earthquake that occurred in 1818 was used as reference. Both earthquakes occurred at a time in which there was no recorded data, and so uncertainties regarding the values of the seismic parameters exist. Therefore, a parametric analysis to very the effect of the seismic design data on the liquefaction potential index PL was carried out. The results obtained for CPT a 1 are shown in Figure 11 and in Table 1. For earthquakes with a magnitude lower than M = 6.5 and with a maximum acceleration lower than amax = 0.5g, the liquefaction risk is low. For more severe earthquakes, the liquefactionriskcan be high, but never very high.
0 5 10 15 Earthquake Resistant Engineering Structures 309 Figure 10: Liquefactim potential inaex ^ tor CKl n. l ^ seismic design data on the liquefaction potential index for CPT a 1 6. Influence of the method used on the results in terms of the liquefaction potential index Although most of the scientic community considers the R&W method for the liquefaction risk analysis from CPT to be the most reliable (Crespellani et al.*), in order to very the influence of the chosen procedure on the results of the analysis, two other recent and reliable methods were applied to the eleven CPT of the La Rena site in Catania. They are the Suzuki et al. method and the Olsen* method. Considering the R&W method as a reference, the results obtained show that the Suzuki et al/ method underestimates the PL, while the Olseri^ method generally slightly overestimates the PL value. In any case, the agreement is very good, confirming the reliability of the methods for assessing the liquefactionriskfrom CPT. M < 5.5 6.0 6.5 7.0 7.5 8.0 8.5 <UO 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0. 08 0.20 0. 30 0. 44 0.03 0.20 032 0.53 0.88 1.26 0.07 0.26 0.49 0.91 1.42 2.15 3.13 0.01 027 0.57 1.20 2.05 3.27 4.41 5.47 022 0.57 1.34 2.59 4.10 5.42 6.67 7.78 0.03 0.43 126 2.79 4.56 6.13 7.55 8.90 10.05 0.18 0.81 224 4.40 626 7.91 9.43 10.65 11.63 0.31 1.40 3.75 5.97 7.93 9.67 11.01 12.06 12.89 Table 1: Effect of seismic design data on the liquefaction potential index
310 Earthquake Resistant Engineering Structures 1 Conclusions The occurrence of liquefaction in SG. La Rena site is strong influenced by expected magnitude and acceleration. According to the Catania Project by the National Group for the Defence Against Earthquake (CNR-GNDT), considering the earthquake scenario of M=7.3 and a,nax=0.35g the liquefaction potential index is 5, which means that the risk could be high, according to the historical documents. Acknowledgement This work has been carried out with the financial support of the National Research Council - National Group for Defence Against Earthquake, contract no.97/00523.pf54. References 1. Cascone E., Castelli F., Grasso S., Maugeri M. (1999) "Zoning for soil liquefaction at Catania city (Sicily)". Proc. ERES99, Catania 15-17 June, 1999. 2. Robertson, P.K., Wride (Fear), C.E. (1997) "Cyclic liquefaction and its evaluation based on SPT and CPT" Final Contribution to the Proc. NCEER Workshop on Liquefaction, Salt Lake City, USA. 3. Suzuki, Y, Koyamada, K., Tokimatsu, K. (1997) "Prediction of liquefaction resistance based on CPT tip resistance and sleeve friction" Proc. XIV ICSMFE, vol. 1, 603-606, Hamburg. 4. Olsen, R.S. Koester J.P. (1998) "Predictionof liquefaction resistance using the CPT", Proc. Int. Symposium on Cone Penetration Testing, CPT'95, Linkoping, Sweden, Vol.2, pp.251-256,1995. 5. Iwasaki, T., Tatsuoka, F., Tokida, K., Yasuda, S. (1978) "A practical method for assessing soil liquefaction potential based on case studies at various sites in Japan" Proc. 2nd Int. Conf. on Microzonation for Safer Construction - Research and Application, San Francisco, Calornia, vol. 2, 885-896. 6. Youd, T.L. (1997) - Proc. of NCEER Workshop on Evaluation of Liquefaction, Salt Lake City, USA, Jan. 1996. 7. Robertson, P.K. (1990) "Soil classication using the CPT" Can. Geotech. J., Ottawa, Canada, Vol. 27, N. 1,151-158. 8. Crespellani T., Madiai C, Vannucchi, G. (1997) - Valutazione del potenziale di liquefazione di vaste aree mediante prove CPT. Atti 8 Convegno Nazionale "L'lngegneria sismica in Italia", Taormina (ME), Settembre 1997.