ACTA GEOTECHNICA SLOVENICA Ustanovitelji Founders Univerza v Mariboru, Fakulteta za gradbeništvo, prometno inženirstvo in arhitekturo University of Maribor, Faculty of Civil Engineering, Transporta­tion Engineering and Architecture Univerza v Ljubljani, Fakulteta za gradbeništvo in geodezijo University of Ljubljana, Faculty of Civil and Geodetic Engineering Univerza v Ljubljani, Naravoslovnotehniška fakulteta University of Ljubljana, Faculty of Natural Sciences and Engineering Slovensko geotehniško društvo Slovenian Geotechnical Society Društvo za podzemne in geotehniške konstrukcije Society for Underground and Geotechnical Constructions Izdajatelj Publisher Univerza v Mariboru, Fakulteta za gradbeništvo, prometno inženirstvo in arhitekturo Faculty of Civil Engineering, Transportation Engineering andArchitecture Odgovorni urednik Editor–in–Chief Borut Macuh University of Maribor Tehni°na urednica Technical Editor Tamara Bra°ko University of Maribor Uredniki Co Editors Jakob Likar Geoportal d.o.o. Janko Logar University of Ljubljana Primož Jeluši° University of Maribor Stanislav Škrabl University of Maribor Goran Vižintin University of Ljubljana Bojan Žlender University of Maribor Posvetovalni uredniki Advisory Editors Heinz Brandl Vienna University of Technology Chandrakant. S. Desai University of Arizona Bojan Majes University of Ljubljana Pedro Seco e Pinto National Laboratory of Civil Eng. Lektor Proof Reader Paul McGuiness Naklada Circulation - Cena Price 25 EUR/letnik - 25 EUR/vol.; (50 EUR for institutions/za institucije) ISSN: 1854-0171 Uredniški odbor Editorial Board Marx Ferdinand Ahlinhan National University in Abomey Amin Barari Aalborg University ~eodoros Hatzigogos Aristotle University of ~essaloniki Vojkan Jovi°i° IRGO-Ljubljana Rolf Katzenbach Technical University Darmstadt Nasser Khalili ~e University of New South Wales, Sydney Svetlana Melentijevic Complutense University of Madrid Seyed Hamed Mirmoradi Federal University of Rio de Janeiro Ana Petkovšek University of Ljubljana Borut Petkovšek Slovenian National Building and Civil Engineering Institute Mihael Ribi°i° University of Ljubljana César Sagaseta University of Cantabria Patrick Selvadurai McGill University Stephan Semprich University of Technology Graz Devendra Narain Singh Indian Institute of Technology, Bombay Abdul-Hamid Soubra University of Nantes Kiichi Suzuki Saitama University Antun Szavits-Nossan University of Zagreb Kosta Urumovi˛ Croatian geological survey Ivan Vani°ek Czech Technical University in Prague Založnik Published by Univerzitetna založba Univerze v Mariboru Slomškov trg 15,2000 Maribor, Slovenija e-pošta: zalozba@um.si,http://press.um.si/,http://journals.um.si/ University of Maribor Press Slomškov trg 15,2000 Maribor, Slovenia e-mail: zalozba@um.si, http://press.um.si/,http://journals.um.si/ Naslov uredništva Address ACTA GEOTECHNICA SLOVENICA Univerza v Mariboru, Fakulteta za gradbeništvo, prometno inženirstvo in arhitekturo Smetanova ulica 17, 2000 Maribor, SlovenijaTelefon / Telephone: +386 (0)2 22 94 300Faks / Fax: +386 (0)2 25 24 179E-pošta / E-mail: ags@um.si Tisk Print Spletni naslov web Address - http://www.fg.uni-mb.si/journal-ags/ Revija redno izhaja dvakrat letno. ^lanki v reviji so recenzirani s stra­ni priznanih mednarodnih strokovnjakov. Baze podatkov v katerih je revija indeksirana: SCIE - Science Citation Index Expanded, JCR – Journal Citation Reports / Science Edition, ICONDA - ~e inter­national Construction database, GeoRef. Izid publikacije je fnan°no podprla Javna agencija za raziskovalno dejavnost Republike Slovenije iz naslova razpisa za sofnanciranje doma°ih periodi°nih publikacij. ~e journal is published twice a year. Papers are peer reviewed by renowned international experts. Indexation data bases of the journal: SCIE - Science Citation Index Expanded, JCR – Journal Citation Reports / Science Edition, ICONDA- ~e international Construction database, GeoRef. ~e publication was fnancially supported by Slovenian Research Agency according to the Tender for co-fnancing of domestic periodicals. T. Sarici et al.: The resilient modulus of hybrid construction and demolition wastes reinforced by a geogrid THE RESILIENT MODULUS OF HYBRID CONSTRUCTION AND DEMOLITION WASTES REINFORCED BY A GEOGRID Talha Sarici (corresponding author) Inonu University, Department of Civil Engineering Malatya, Turkey E-mail: talha.sarici@inonu.edu.tr Aykan Mert General Directorate of Highways, Research and Development Department Ankara, Turkey E-mail: amert@kgm.gov.tr MODUL PROŽNOSTI HIBRIDNIH GRADBENIH ODPADKOV IN ODPADKOV PRI RUŠENJU,OJA°AN Z GEOMREŽO Bahadir Ok Adana Alparslan Turkes Science and Technology University, Department of Civil Engineering Adana, Turkey E-mail: bahadirok@atu.edu.tr Senol Comez General Directorate of Highways, Research and Development Department Ankara, Turkey E-mail: scomez@kgm.gov.tr https://doi.org/10.18690/actageotechslov.19.2.2-14.2022 construction and demolition waste, Geogrid, geotechni-gradbeni odpadki in odpadki pri rušenju objektov, cal engineering, sustainability, resilient modulus, waste geomreže, geotehni°ni inženiring, trajnost, modul management prožnosti, ravnanje z odpadki ˛e use of construction and demolition wastes (C&D) in engineering applications is an important development for better sustainability. ˛e main objective of this study, therefore, was to increase the use of C&D by improving their engineering behaviour. For this purpose, two methods were employed in this study: frst, adding the virgin aggre­gates (VA) to the C&D, called hybrid C&D (C&D-VA), and second, reinforcing the C&D with a geogrid material. Test samples were prepared in six groups. ˛e frst three test groups were prepared with C&D,VA and C&D-VA. ˛e other three test groups were formed with geogrid-rein­forced C&D,VA and C&D-VA. Firstly, for the strength characteristics of the samples, the unconfned compressive strength and the California bearing ratio values were obtained with large-scale experiments. Subsequently, for the resilient behaviour of the samples, the resilient modu­lus values were determined using a large-scale triaxial test device. Consequently, some signifcant improvements Uporaba gradbenih odpadkov in odpadkov pri rušenju (C&D) v inženirskih objektih je pomemben pri razvoju za ve°jo trajnost. Glavni cilj pri°ujo°e študije je torej pove°ati uporabo C&D z izboljšanjem njihovega inženirskega obnašanja. V ta namen sta bili v tej študiji uporabljeni dve metodi, in sicer prva dodajanje neobdelanih agregatov (VA) v C&D, imenovano hibridni C&D (C&D-VA), in druga, oja°itev z geomrežami C&D. Preizkušanci so bili pripravljeni v šestih skupinah. Prve tri preizkusne skupine so bile pripravljene s C&D,VA in C&D-VA. Druge tri preizkusne skupine so bile pripravljene tako, da so zgor­njim trem skupinam bile dodane geomreže, torej oja°ane C&D,VA in C&D-VA. Najprej so bile z izvajanjem obsežnih preizkusov enoosne tla°ne trdnosti in kalifornij­skega faktorja nosilnosti pridobljene vrednosti trdnostnih karakteristik preizkušancev. Nato so bile za deformacijsko obnašanje vzorcev dolo°ene vrednosti modula prožnosti z uporabo velike triosne preskusne naprave. Posledi°no 2. Acta Geotechnica Slovenica, 2022/2 T. Sarici et al.: The resilient modulus of hybrid construction and demolition wastes reinforced by a geogrid were achieved via the methods employed in this study. In addition, it was observed that the best reinforcement e˙ect for the C&D occurred when the geogrid was used and the VA was added to the C&D. 1 INTRODUCTION Construction and demolition wastes (C&D) occur in construction, repair, maintenance, environmental disasters and demolition activities [1]. ~e C&D can consist of di˘erent types of materials, depending on the construction or demolition activities. ~ese materials can be concrete, brick, tile, ceramic, wood, glass, plastic, bituminous mixtures, coal, petroleum products, metals, soil pieces, insulating materials, building materials containing asbestos, gypsum-based construction materials, etc. [2]. On the other hand, C&D is some of the heaviest and most voluminous waste and constitutes between 30 % and 40 % of the total solid waste [3]. ~erefore, this solid waste can cause negative impacts on the environment when it is stored in a landfll. In addi­tion, this storage is not economic. However, if the C&D is reused in some construction applications by recycling, the storage costs of the solid waste produced by the construction industry can be reduced, the need for the area of the landfll can be diminished, the use of natural resources for construction can be decreased, energy waste and greenhouse-gas emissions can be reduced, and sustainability can be increased [4-6]. Many undeveloped and developing countries store C&D without recycling in landflls. Although some developed countries recycle a part of the C&D, the level of recycling is insuycient [3]. Overall, the recycling of C&D in developed or developing countries should be increased, since recycling, recovery and sustainability are indispensable for our world at this time. ~erefore, many researchers have recommended increased studies on the subject to expand the areas of use for C&D [5-11]. Generally, it is predicted that the C&D in some geotechnical applications such as fllings for various aims and base/subbase layer in an unbound pavement can be reused instead of virgin aggregate. Accordingly, several studies were conducted involving conventional laboratory tests such as proctor, unconfned compressive strength, and California bearing ratio tests. In these studies, it was mentioned that the C&D could be a good alternative to virgin aggregates in fllings. However, in many studies, it is stated that the quality of the C&D in terms of several geotechnical and physical parameters is less than that of virgin aggregates [2, 8]. In these stud­ies, it is suggested that the engineering behaviours of je bilo z metodami, uporabljenimi v tej študiji, doseženih nekaj pomembnih izboljšav. Poleg tega je bilo ugotovljeno, da je najboljši u°inek oja°itve za C&D dosežen, ko je bila nameš°ena geomreža in dodana VA v C&D. the C&D should be improved. ~erefore, some studies have used geosynthetics [12, 13] or additives [9, 14, 15] to improve the engineering properties of the C&D. Moreover, for the same purpose, the C&D was mixed with virgin aggregates in a few studies, and subsequently some tests were carried out with mixture aggregates [16-18]. However, because the studies usually involve small-scale conventional laboratory tests, more compre­hensive research is needed to increase the reuse of C&D. In some geotechnical applications, such as fllings and unbound pavement layers, granular soils are generally used as the flling material. Geogrids, which are a type of geosynthetics and used for reinforcement purposes, can be more suitable for an improvement of the C&D since they have an interlocking mechanism with particles if the C&D is to be transformed into a granular material and reused. Although in the literature there are many studies on the advantages of the interaction mechanism between geogrids and virgin granular soils [19-28], there are only a few studies related to C&D reinforced by geogrids [12, 13]. In addition, in studies on C&D reinforced by a geogrid, it was emphasized that the subject should be examined in more detail. On the other hand, the granular fll layers should be capable of resisting static and repeated stresses [29, 30]. Generally, the C&D consists of a wide variety of waste materials that would further complicate the behaviour under repeated stresses. When the use of C&D instead of virgin aggregates is investigated, the complex behaviour under repeated stresses of the C&D needs to be determined. ~e resilient modulus tests, which are performed by applying the cycling stresses in di˘erent stress combina­tions in a repeated load triaxial test device, can us help to understand this complex behaviour. For this reason, a few researchers have conducted resilient modulus tests on C&D. In those studies, it was stated that the C&D needs reinforcement in terms of resilient behaviour [31-34]. Recently, Chen et al. [29] in their study stated that the resilient behaviour of low-quality virgin aggre­gate can be improved with a geogrid. Accordingly, the resilient behaviour of the C&D can be improved with a geogrid, for example, low-quality aggregates. In this regard, Rahman et al. [35] in their study reported that the inclusion of a geogrid had e˘ects on the resilient modulus and permanent deformation behaviour of the C&D. However, in this study, it was mentioned that studies that are related to C&D reinforced with geogrids T. Sarici et al.: The resilient modulus of hybrid construction and demolition wastes reinforced by a geogrid are limited and the reinforcement e˘ects of the geogrids under repeated stresses are still unknown. In summary, it is necessary to improve some properties such as the resilient behaviour and the compressive strength of the C&D to increase the reuse of C&D in fllings. However, according to the literature, further investigations need to be conducted for that. In this study, two di˘erent improvement methods were inves­tigated to improve the C&D. ~e frst was to mix the C&D with virgin aggregates (VA). In other words, to produce a type of hybrid C&D (C&D-VA), which is an easy improvement method. ~e second was reinforcing the C&D with a geogrid, which is a widely used method to improve granular soils. In addition, in the case of using both methods together, the improvement of the C&D was investigated as well. ~e e˘ects of improvements were evaluated for both monotonically increased stresses and cycling stresses. Accordingly, tests were carried out on large-scale test samples. It is believed that this study will make a great contribution to the literature as it investigates the e˘ects of di˘erent improvement types and evaluates these e˘ects in terms of monotonically increased stresses and cycling stresses. It is also estimated that the suggestions to be presented to the designer at the end of the study will increase the reuse rate of the C&D. Besides, this study o˘ers alternatives related to reusing even low-strength C&Ds by means of some improvements. ~e focus of this study is to investigate the reusability by improving the C&D obtained from low-strength, classically reinforced concrete structures, of which the mean compressive strength of the concrete core samples was 14.5 MPa. ~e C&D was obtained by carrying out several recycling processes. For the improve­ment of the C&D, two methods were used for the mixing with the VA of the C&D, (C&D-VA), and the reinforcing with the geogrid. In addition, those methods were also used together . Reinforced and unreinforced C&D and obtained from debris as recycled aggregates, virgin aggregates (VA) taken from a quarry and hybrid aggre-gates (C&D-VA) derived by mixing the C&D with the VA in equal amounts (Figure 1). ~e debris was taken from the group of low-strength RC structures, where the concrete compressive strengths of the core samples were varied between 7.5 MPa and 20 MPa, and the C&D was obtained by carrying out several recycling processes on this debris. Firstly, the debris was transferred to a crushing machine to produce proper-sized granular materials. Subsequently, steel bars and iron pieces in the debris were removed by passing them through a magnet system. Aer this process, the debris was crushed and the C&D materials, which have three di˘erent grain-size ranges, i.e., 0–5, 5–12, and 12–25 mm [36], were obtained. Subsequently, based on the particle size of those C&D materials, a mixture calcu­lation was made to obtain a gradation that is suitable for use in highway base and sub-base courses [37]. Finally, the C&D materials obtained in di˘erent gradations were mixed and the C&D used as a test sample was obtained. On the other hand, for a comparison with the C&D, the VA with limestone particles was obtained from a quarry in Turkey. ~e gradation of the VA was made suitable to Figure 2. Gradations of the C&D, VA and C&D-VA with limit values recommended by ASTM [37]. 4. Acta Geotechnica Slovenica, 2022/2 T. Sarici et al.: The resilient modulus of hybrid construction and demolition wastes reinforced by a geogrid use in highway base and sub-base courses [37], similar to the C&D [38]. Furthermore, the C&D with the VA, which has similar gradations, was mixed in equal amounts and the C&D-VA mixtures were obtained. ~e gradations of the C&D, the VA and the C&D-VA are shown in Figure 2. When the C&D was examined in detail, it was clear that the C&D includes di˘erent recycled wastes, such as concrete, aggregate, brick, glass and some other materi­als. According to the tests carried out considering BS EN 933-11 [39], the C&D in this study consists of 36.33 % concrete (Rc), 52.65 % aggregate (Ru), 10.53 % brick (Rb), 0.11 % glass (Rg) and 0.38 % other materials (metals, non-foating wood plastic, rubber, plaster) (X). It also contains 0.7 kN/m3 of foating particles (FL) [8]. Some physical and geotechnical properties of the C&D, the VA and the C&D-VA were obtained with laboratory tests, such as a sieve analysis, fatness index, Los Angeles abrasion, water absorption, pycnometer tests and modi­fed compaction tests [36, 40-44]. ~e results obtained from these tests are shown in Table 1. In addition, compaction curves obtained from modifed compaction Figure 3. Compaction curves of the granular materials. tests are shown in Figure 3. Detailed characteristics of the C&D and the VA were reported by Ok et al. [8]. 2.2 Geogrid In this research, a triaxial geogrid, which is obtained from a manufacturer, was used to improve the resilient modulus and the unconfned compressive strength of the C&D and the C&D-VA. ~is triaxial geogrid was manu­factured from punched polypropylene sheets and has an equilateral direction to form its triangular apertures. ~e texture of the geogrid is shown in Figure 4, and the physical and mechanical properties of the geogrid, as provided by the manufacturer, are presented in Table 2. Table 2. Properties of geogrid. Properties Unit Description or value Raw Material - Polypropylene Aperture Type - Triangle Aperture Dimensions mm 40×40×40 ~ickness mm 1.1 Tensile Strength at 5 % strain, md/cmd* kN/m 300 *: machine direction/cross machine direction Table 1. Physical and geotechnical properties of C&D, VA and C&D-VA mixture. Properties Unit C&D VA C&D-VA Coeycient of uniformity (Cu) - 41.87 35.88 39.97 Coeycient of curvature (Cc) - 1.06 1.89 1.35 Flakiness index % 11.68 12.66 12.11 Los Angeles abrasion loss % 33.58 23.40 29.89 Particle Density (^s) kN/m3 26.30f, 26.10c 26.90f, 27.10c 26.55f, 26.50c Water absorption % 6.82f, 4.06c 0.40f, 0.36c 3.88f 2.51c Maximum dry unit weight (^drymax) kN/m3 20.77 23.90 21.10 Optimum water content, (wopt) % 9.7 6.0 8.5 f: Fine particle, c: Coarse particle T. Sarici et al.: The resilient modulus of hybrid construction and demolition wastes reinforced by a geogrid 3 TESTING METHODS 3.1 California bearing ratio (CBR) ~e CBR test is commonly used to compare the strength of flling materials. ~e test is performed by penetrating a cylindrical steel piston of 50 mm diameter into the sample, which is placed in a mold (152.4 mm diameter), at a rate of 1.27 mm/min [45]. ~e result of the test can be presented both in terms of load–displacement curves and percent relative (CBR value) to the reference value in the ASTM D1883–99 [45]. For the CBR tests, the flling materials with optimum water content were placed in a mold, which is used in modifed compaction tests, by compacting to provide their maximum dry unit weight. ~en, the prepared samples were tested accord­ing to ASTM D1883–99 [45] and their CBR values were obtained. 3.2 Large-scale unconfned compressive strength (UCS) ~e UCS test is one of the most popular tests used as a key design index parameter for estimating the sti˘ness of soils. ~e UCS test includes the application of an axial vertical load through loading platens, using strain-control conditions, to a cylindrical soil sample that is unconfned. ~e maximum unit stress obtained from the result of the UCS test is defned as the UCS [46]. Large-scale UCS tests on geogrid-reinforced and unreinforced C&D, VA and C&D-VA were performed in this study. In the preparation of test samples, since the maximum aggregate size of the flling materials is 20 mm, a large-scale split mold, in which the e˘ective internal height is 300 mm and the e˘ective internal diameter is 150 mm, was used. For large-scale UCS tests, the flling materials with optimum water content were placed in a large-scale split mold and compacted to achieve the maximum dry unit weight. According to ASTM D2166 [46], the UCS tests were conducted by applying an axial strain rate of 0.5 % per minute to the samples. 3.3 Resilient modulus (MR) ~e fllings beneath oil storage tanks, silos or machine foundations and embankments such as road base/ sub-base are subjected to repeated loads. In these cases, and many similar situations, the resilient behaviour of the fllings is signifcant in addition to the unconfned compressive strength. However, the resilient behaviour of granular materials depends on some agents. For example, granular materials can have di˘erent resilient deformation values according to the stress levels applied to them. Hence, the resilient behaviour can usually be characterized by the resilient modulus (MR), which has di˘erent values for di˘erent stress levels. Accordingly, MR was used by several researchers to characterize the resilient behaviour of the base/sub-base course material and the subgrade soil [47]. MR is defned as the ratio of the deviator stress to the vertical elastic deformation [48]. In this study, MR tests were performed using a large-scale cyclic triaxial test device to determine the resilient behaviour of the geogrid-reinforced and unreinforced C&D, VA and C&D-VA [49]. ~e MR test samples with optimum water content were placed in the split mold with a diameter of 150 mm and a height of 300 mm, by providing the maximum dry unit weight. MR tests were performed with 1000 cycles in the initial stage and then 100 cycles in each stage, for a total of 2500 load cycles. Since permanent deformation values are almost constant in the last cycles of the stress stages, the resilient modulus value of any stress stage is determined by considering the last fve cycles. ~e stress stages for aggregate materials are shown in Table 3. According to AASHTO T-307 [49], the load pulses applied in MR tests had a haversine-shaped loading of 0.1 seconds and rest periods of 0.9 seconds. Table 3. Stress stages and values according to AASHTO [49]. Stress stages Confning stress (kPa) Deviator stress (kPa) Bulk stress (kPa) Stress stages Confning stress (kPa) Deviator stress (kPa) Bulk stress (kPa) 0 103.4 103.4 413.7 8 68.9 137.9 344.7 1 20.7 20.7 82.7 9 68.9 206.8 413.7 2 20.7 41.4 103.4 10 103.4 68.9 379.2 3 20.7 62.1 124.1 11 103.4 103.4 413.7 4 34.5 34.5 137.9 12 103.4 206.8 517.1 5 34.5 68.9 172.4 13 137.9 103.4 517.1 6 34.5 103.4 206.8 14 137.9 137.9 551.6 7 68.9 68.9 275.8 15 137.9 275.8 689.5 6. Acta Geotechnica Slovenica, 2022/2 T. Sarici et al.: The resilient modulus of hybrid construction and demolition wastes reinforced by a geogrid 3.4 Geogrid reinforcement Geogrid reinforcement (RF) has been used to improve the resilient behaviour and compressive strength of flling materials (C&D and C&D-VA) obtained from recycled aggregates in this study. Furthermore, to compare the e˘ect of geogrid reinforcement, samples of the geogrid-reinforced VA were also prepared and tested. For this purpose, the MR and the UCS tests were carried out on geogrid-reinforced and unreinforced C&D, VA and C&D-VA. Abu-Farsakh et al. [25] conducted resilient modulus tests on virgin aggregates by placing the geogrid at di˘erent locations on the test specimen. Consequently, they stated that when the geogrid is placed in the upper or middle of the test specimen, more improvement than other locations was obtained. In this study, considering some studies in the literature [25, 35, 50], the geogrid reinforcement was placed at the mid-height of the test samples. ~e placement and layout of the geogrid reinforcement are shown in Figure 5. 4 DISCUSSION OF RESULTS 4.1 Evaluation of the CBR values of flling aggregates ~e CBR tests were carried out on the C&D, the VA and the C&D-VA samples. ~e CBR values and load-displace­ment curves of the samples were determined according to the results of those tests. ~e CBR values of the C&D, the VA and the C&D-VA samples were 98.99 %, 125.16 % and 105.51 %, respectively. ~e load–displacement (N-s) curves of those aggregates are shown in Figure 6. Figure 6. Load–displacement curves of the fllingaggregates in CBR tests. Considering the CBR test results, the behaviour of the load-displacement curves for all the samples were similar for displacements of less than 2 cm. However, as with the displacement increases, the situation changed in favour of the VA. ~is result is attributed to the fact that the VA particles are stronger than the C&D particles, as seen in the Los Angeles abrasion tests [8]. However, the CBR values indicate that the C&D and the C&D-VA samples are appropriate for use as a flling material according to some technical specifcations [51, 52]. 4.2 Comparison C&D with VA Firstly, the MR and the UCS tests were performed to determine the resilient behaviour and compressive strength of the C&D. Subsequently, for comparison, those tests were conducted on the VA. ~e results of those tests are shown by comparing each in Figure 7. Figure 7 shows that the C&D is less than that of the VA in terms of both MR and UCS. ~e UCS value of the VA is 30.7 % higher than that of the C&D. Moreover, Figure 5. Placement and layout of the geogrid reinforcement. T. Sarici et al.: The resilient modulus of hybrid construction and demolition wastes reinforced by a geogrid Figure 7. Stress-displacement curves and the MR values of the C&D and the VA. for all the stress stages, the MR values of the C&D are lower than those of the VA. Although it has been stated in various studies that C&D can be used in some fllings, even with this performance, it has been mentioned in those studies that various improvements are needed to increase the performance of the C&D [6, 8, 16, 35, 53, 54]. ~erefore, in this study, the performance of the C&D was increased by mixing the C&D with VA or using the geogrid reinforcement. 4.3 Evaluation of the C&D-VA mixture A new aggregate mixture, namely the hybrid C&D (C&D-VA), was obtained by mixing the C&D with VA in the same proportions to increase the MR and the UCS of the C&D. ~e results of the UCS and the MR tests of the C&D-VA are shown by comparing with that of the C&D and the VA in Figure 8. Mixing the C&D with the VA increased the UCS by approximately 11 %. Also, in all stress stages, the MR values of the C&D-VA are more than those of the C&D. However, the improvement of both the MR and the UCS are very limited, and the MR and the UCS values of the VA are greater than the C&D-VA. Even if according to those results, also the C&D-VA like the C&D may be an alternative to the VA to use as a flling material, it is thought that it might need an improvement such as geogrid reinforcement [6, 8, 35]. 4.4 Effects of geogrid reinforcement Geogrid reinforcement (RF) was used to increase the MR and the UCS of the C&D and C&D-VA. Furthermore, to compare the e˘ect of geogrid reinforcement, the MR and the UCS tests were also carried out on the VA reinforced by the geogrid. Accordingly, the e˘ects of geogrid on 8. Acta Geotechnica Slovenica, 2022/2 T. Sarici et al.: The resilient modulus of hybrid construction and demolition wastes reinforced by a geogrid Mean Bulk Stress (kPa) Figure 9. Stress-displacement curve and the MR values of the C&D (RF) in comparison with the C&D and the VA. those parameters were discussed in terms of the C&D and the VA according to the results of the tests on both reinforced C&D and reinforced VA. ~e results of the UCS and the MR tests of the geogrid reinforced C&D, namely, C&D (RF), are shown by comparing with that of the C&D and the VA in Figure 9. According to the results of tests performed on the C&D (RF), the UCS value of the C&D (RF) was approximately 35 % higher than that of the C&D. In other words when the C&D is reinforced by a geogrid, the UCS value exceeded that of the VA. However, in all the stress stages, although the MR values of the C&D (RF) are more than those of the C&D, they are less than those of the VA. ~erefore, there is a signifcant improvement in mono­tonic stress for the geogrid-reinforced C&D, while the improvement is limited in cycling stress. Consequently, for fllings exposed to static loads, the geogrid-reinforced C&D can achieve the performance of natural aggregates, but it may be necessary to develop di˘erent solutions to obtain the performance of natural aggregates in fllings exposed to repeated stress such as cycling stress. For this, reinforcement of the C&D-VA sample with a geogrid was considered. Accordingly, the UCS and the MR tests on the C&D-VA reinforced by the geogrid, namely C&D-VA (RF), were performed. ~e results of those tests are shown by comparing with that of the C&D and the VA in Figure 10. According to the results of the tests performed on the C&D-VA (RF), the UCS value of the C&D-VA (RF) was obtained as approximately 44 % and 10 % higher than that of the C&D and that of the VA, respectively. On the other hand, for all the stress stages, the MR values of the C&D-VA (RF) are more than those of the C&D. Moreo­ver, in the low-stress stages, although the MR values of the C&D-VA (RF) are slightly less than those of the VA, in high-stress stages, the MR values of the C&D-VA (RF) Figure 10. Stress-displacement curve and the MR values of the C&D-VA (RF) in comparison with the C&D and the VA. T. Sarici et al.: The resilient modulus of hybrid construction and demolition wastes reinforced by a geogrid are close to those of the VA. ~is result is thought to be obtained due to the geogrid's reinforcement mechanisms. Geogrids have main reinforcement functions, such as lateral confnement and a membrane e˘ect. [55]. ~e lateral confnement function, one of the geogrid reinforcement mechanisms, is due to the soil particles interlocking within the geogrid aperture. While soil particles cannot resist the tensile stress, the geogrid mate­rial can resist a higher tensile stress than soil particles. As the soil particles begin to deform laterally, they fall into the geogrid apertures. ~is situation caused the interlock­ing mechanism. ~us, the tensile stresses occuring in the soil particles transmit to the geogrid. Since the geogrid can resist much more tensile stress, the strength of the soil layer increases [48]. ~e membrane e˘ect, another of the geogrid-reinforcement mechanisms, occurs as a result of the deformation of the soil. When any stress is applied to the soil layers, the soil layers can move down from its current position. As a result of this situation, the geogrid is deformed and tensioned. ~e vertical deformation creates a concave shape in the geogrid. Due to tensile sti˘ness of the geogrid, the concave shape performs an upward force to support the applied load and reduce the vertical stress on the soil layers. However, to achieve this e˘ect, there must be a signifcant deforma­tion [56]. When Figure 10 is examined, the deformation and stress increase, the improvement of the sample increases due to the reinforcement mechanisms, such as the membrane e˘ect and the lateral confnement of the geogrid. Similarly, the same reinforcement mechanism was observed in geogrid-reinforced (i.e., the VA (RF)) and unreinforced VA. ~e results of the MR and the UCS tests of the VA (RF) and VA are shown Figure 11. As the deformation and stress increase, the improvement of the sample increases. For virgin aggregates, this event is in line with previous studies in the literature. In this study, in the geogrid-reinforced C&D, a reinforcement mechanism similar to the geogrid-reinforced VA was observed. ~erefore, it was considered that C&D is a convenient material to reinforce with a geogrid. However, it should be considered that the reinforcement with a geogrid is more e˘ective in high deformation and stress. 4.5 UCSR and MRR Two coeycients, the UCSR and the MRR, have been defned as dimensionless parameters obtained from the results of tests. ~e UCSR was defned as the ratio of the UCS value obtained in the result of a test, which will be compared to the UCS value of the C&D. Similarly, MRR was defned as the ratio of the MR value obtained in the result of a test, which will be compared to the MR value of the C&D, which has the same stress stages. A calculation of those coeycients is shown in Equation 1 and 2. ~e UCSR and MRR values calculated from the test results are shown in Figure 12 and Figure 13, respectively. (1) (2) As seen in the UCSR values obtained from the results of the tests, for monotonically increased stresses, the performance of the C&D can increase suyciently to obtain that of the VA when it is reinforced by a geogrid only. However, as seen in the MRR values, in repeated stresses, the reinforcing with the geogrid of the C&D might not be enough to obtain the performance of the VA. In this case, i.e., under repeated stresses, if the C&D is mixed with the VA and then the mixture is reinforced 10. Acta Geotechnica Slovenica, 2022/2 T. Sarici et al.: The resilient modulus of hybrid construction and demolition wastes reinforced by a geogrid by the geogrid, it is clear that the performance of the VA can be achieved. It was considered that the reason for this was that the e˘ect of cyclic loads on brittle soil grains could be greater. 5 CONCLUSIONS In this study, laboratory tests such as resilient modulus tests and unconfned compressive tests, including large-scale tests, to improve the resilient behaviour and compressive strength of the C&D were performed. For this purpose, the e˘ectiveness of some improvement methods, such as both the mixing with the VA of the C&D (in other words producing a type of hybrid C&D) and reinforcing the C&D with geogrid was evaluated. On the basis of the results of these tests, the following conclusions can be drawn:  ~e UCS value of the C&D was obtained as 30.7 % less than that of the VA. Moreover, it was seen that the MR values in all the stress stages of the C&D are less than those of the VA. ~ese results, similar to those from Los Angeles abrasion tests, are assumed to be due to the VA particles being stronger than the C&D particles. ~e CBR test results confrm this result. So, the test results show that there is a quality di˘erence between the C&D and the VA in terms of both monotonically increasing and repeated stresses.  According to the results of tests on the hybrid aggre­gate (C&D-VA), the MR values in all the stresses stages and the UCS value of the C&D-VA were more than those of the C&D. In the case of adding the VA to the C&D, the UCS value was increased by 11 %. However, the improvement is limited and the values of C&D-VA do not reach those of the VA.  C&D (RF)’s UCS was approximately 35 % higher than that of C&D, thus exceeding that of the VA. However, it was found that although the MR values of the C&D (RF) are more than those of the C&D, they are less than those of the VA for all stress stages. ~erefore, it was a signifcant improvement in mono­tonically increased stresses, while the improvement was limited in the cycling stresses because the C&D is reinforced by a geogrid. If C&D is to be used in the construction of a fll, these consequences should be considered for a flling material that could be subjec­ted to repeated stresses.  According to the results of tests on the hybrid aggregate reinforced by a geogrid, the UCS value of the C&D-VA (RF) is approximately 44 % higher than that of the C&D, and the MR values of the C&D-VA (RF) are more than those of the C&D.  ~e UCS value of the C&D-VA (RF) was approxi­mately 10 % higher than that of the VA. Also, the MR values of the C&D-VA (RF) are close to those of the VA in high-stress stages, although in the low-stress stages they are slightly less. ~is result is thought to be due to reinforcement mechanisms, such as lateral confnement and the membrane e˘ect of the geogrid.  ~e reinforcement mechanisms of all the test samples reinforced with the geogrid were similar. ~erefore, the C&D could be a suitable material to reinforce with a geogrid.  In the case of both mixing with the VA and reinfor­cing with the geogrid, for the C&D it can be consi­dered that the best improvement was achieved on both the monotonically increased and the repeated stresses. With these improvements it can be possible to have durable fllings even when using low-strength C&D, and this can increase the reuse of the C&D. Nevertheless, it should be considered that reinfor­cement with a geogrid is more e˘ective for high deformation and stress in designs.  Due to the energy-absorption feature of the geogrid, there are important advantages in dynamic cases. So, T. Sarici et al.: The resilient modulus of hybrid construction and demolition wastes reinforced by a geogrid it is recommended to conducted studies that include earthquake analysis such as Edinçliler and Yildiz [57] and Yildiz [58] for a better understanding of the behaviour of geogrid-reinforced C&D and C&D-VA. REFERENCES [1] Shen, L.Y., Tam, V.W., Tam, C.M., Drew, D. 2004. Mapping approach for examining waste management on construction sites. Journal of construction engineering and management 130(4): 472-481, https://doi.org/10.1061/(ASCE)0733­9364(2004)130:4(472). [2] Vieira, C.S., Pereira, P.M. 2015. Use of recycled construction and demolition materials in geotech­nical applications: A review. Resources, Conserva­tion and Recycling 103: 192–204. http://dx.doi. org/10.1016/j.resconrec.2015.07.023. [3] European Commission (DG ENV) 2011. Service contract on management of construction and demolition waste prepared by Bio Intelligent Service, Available online: http://ec.europa.eu/envi­ronment/waste/pdf/2011_CDW_Report.pdf. [4] Arulrajah, A., Piratheepan, J., Aatheesan, T., Bo, M.W.2011. Geotechnical properties of recycled crushed brick in pavement applications. Journal of Materials in Civil Engineering 23(10):1444–52, https://doi.org/10.1061/(ASCE)MT.1943­5533.0000319. [5] Jimenez, J.R., Ayuso, J., Agrela, F., Lpez, M., Galvín, A.P.2012. Utilisation of unbound recycled aggre­gates from selected CDW in unpaved rural roads. Resources, Conservation and Recycling 58: 88–97, https://doi.org/10.1016/j.resconrec.2011.10.012. [6] Jallu, M., Arulrajah, A., Saride, S., Evans, R. 2020. Flexural fatigue behavior of fy ash geopoly-mer stabilized-geogrid reinforced RAP bases. Construction and Building Materials 254-119263: 1-9, https://doi.org/10.1016/j.conbuild­mat.2020.119263. [7] Evangelista, L., de Brito, J. 2014. Concrete with fne recycled aggregates: a review, European Journal of Environmental and Civil Engineering, 18(2), 129-172, https://doi.org/10.1080/19648189.2013.851038 [8] Ok, B., Sarici, T., Talaslioglu, T., Yildiz, A. 2020. Geotechnical properties of recycled construction and demolition materials for flling applications. Transportation Geotechnics 24-100380: 1-9, https://doi.org/10.1016/j.trgeo.2020.100380. [9] Mohammadinia, A., Arulrajah, A., Horpibulsuk, S., Shourijeh, P.T. 2019. Impact of potassium cations on the light chemical stabilization of construction and demolition wastes. Construction and Building Materials 203: 69-74, https://doi.org/10.1016/j. conbuildmat.2019.01.083. [10] Asprone, D., Bilotta, E., Capasso, I., Caputo, D., Flora, A., Liguori, B., Lirer, S. 2015. Re-use of construction and demolition waste for geotechni-cal applications. Geotechnical Engineering for Infrastructure and Development 1(1): 2589–2594. [11] Aldemir, A., Akduman, S., Kocaer, O., Aktepe, R., Sahmaran, M., Yildirim, G., ... Ashour, A. 2022. Shear behaviour of reinforced construction and demolition waste-based geopolymer concrete beams. Journal of Building Engineering, 47, 103861. https://doi.org/10.1016/j.jobe.2021.103861 [12] Vieira, C.S., Pereira, P.M. 2015. Damage induced by recycled construction and demolition wastes on the short-term tensile behaviour of two geosyn­thetics. Transportation Geotechnics 4: 64-75, http://dx.doi.org/10.1016/j.trgeo.2015.07.002. [13] Arulrajah, A., Rahman, M.A., Piratheepan, J., Bo, M.W., Imteaz, M.A. 2013. Interface shear strength testing of geogrid-reinforced construc­tion and demolition materials. Advances in Civil Engineering Materials 2(1): 189–200, https://doi. org/10.1520/ACEM20120055. [14] Arulrajah, A., Mohammadinia, A., Phummiphan, I., Horpibulsuk, S., Samingthong, W. 2016. Stabiliza­tion of recycled demolition aggregates by geopoly­mers comprising calcium carbide residue, fy ash and slag precursors. Construction and Building Materials 114: 864-873, https://doi.org/10.1016/j. conbuildmat.2016.03.150. [15] Li, L., Zhang, H., Xiao, H., Pei, Y., Wang, J. 2020. Mechanical and microscopic properties of alkali-activated fy-ash-stabilised construction and demolition waste. European Journal of Environ­mental and Civil Engineering, 1-17, https://doi. org/10.1080/19648189.2020.1792351 [16] Bennert, T., Papp. Jr. W.J., Maher, A., Gucun-ski, N. 2000. Utilization of construction and demolition debris under trayc-type loading in base and subbase applications. Transporta­tion research record 1714(1): 33-39, https://doi. org/10.3141/1714-05. [17] Ayan, V., Limbachiya, M.C., Omer, JR., Azadani, S.M.N.2014. Compaction assessment of recycled aggregates for use in unbound subbase application. Journal of Civil Engineering and Management 20(2): 169-174, https://doi.org/10.3846/13923730.2 013.801882. [18] Li, L., Qin, L., Xiao, H., Hu, Z., Xu, G., Ma, Q. 2020. ~e triaxial test of construction and demolition (C&D) materials with di˘erent particle sizes and sand contents. European Journal of Environmental 12. Acta Geotechnica Slovenica, 2022/2 T. Sarici et al.: The resilient modulus of hybrid construction and demolition wastes reinforced by a geogrid and Civil Engineering, 1-22, https://doi.org/10.108 0/19648189.2020.1820908. [19] Haas, R., Walls, J., Carroll, R.G. 1988. Geogrid rein­forcement of granular bases in fexible pavements. Transportation research record 1188(1): 19–27. [20] Al-Qadi, IL., Brandon, T.L., Valentine, R.J., Lacina, B.A., Smith, T.E. 1994. Laboratory evaluation of geosynthetic reinforced pavement sections. Trans­portation Research Record. 1439: 25–31. [21] Perkins, S.W. 2002. Evaluation of geosynthetic reinforced fexible pavement systems using two pavement test facilities. Department of Transporta­tion, Federal Highway Administration, Washing­ton, D.C. FHWA/MT-02–008/20040, pp. 1-120, Available online: https://rosap.ntl.bts.gov/view/ dot/42085 [22] Berg, R.R., Christopher, B.R., Perkins, S.W. 2000. Geosynthetic reinforcement of the aggregate base course of fexible pavement structures. GMA White Paper II, Geosynthetic Materials Associa­tion, Roseville, MN, USA, pp. 1-176, Available online: https://geosynthetics.ifai.com/wp-content/ uploads/sites/10/2016/06/geosynthetic_reinforce­ment_gma_white_paper_ii.pdf [23] Cancelli, A. Montanelli, F. 1999. In-ground test for geosynthetic reinforced fexible paved roads. In Twelh European Conference on Soil Mechanics and Geotechnical Engineering. Roseville, MN, vol. 2, pp. 863–878. [24] Nazzal, M., Abu-Farsakh, M., Mohammad, L. 2007. Laboratory characterization of reinforced crushed limestone under monotonic and cyclic loading. Journal of Materials in Civil Engineering 19(9): 772-783, https://doi.org/10.1061/(ASCE)0899­1561(2007)19:9(772). [25] Abu-Farsakh, M., Souci, G., Voyiadjis, G.Z., Chen, Q.2012. Evaluation of factors a˘ecting the perfor­mance of geogrid-reinforced granular base mate­rial using repeated load triaxial tests. Journal of Materials in Civil Engineering 24(1): 72-83, https:// doi.org/10.1061/(ASCE)MT.1943-5533.0000349. [26] Tutumluer, E., Huang, H., Bian, X. 2012. Geogrid-aggregate interlock mechanism investigated through aggregate imaging-based discrete element modeling approach. International Journal of Geomechanics 12(4): 391-398, https://doi. org/10.1061/(ASCE)GM.1943-5622.0000113. [27] Kamalzare, M., Ziaie-Moayed, R. 2011. Infuence of geosynthetic reinforcement on the shear strength characteristics of two-layer sub-grade. Acta Geotechnica Slovenica, 8(1), 39-49. [28] Haimin, W., Yiming, S., Linjun, D., Zhaoming, T., 2015. Implementation and verifcation of a geosynthetic-soil interface constitutive model in the Geogrid element of FLAC3D. Acta geotechnica Slovenica, 12(1), 27-35. [29] Chen, Q., Abu-Farsakh, M., Voyiadjis, G.Z., Souci, G.2013. Shakedown analysis of geogrid-reinforced granular base material. Journal of materials in civil engineering 25(3): 337-346, https://doi. org/10.1061/(ASCE)MT.1943-5533.0000601. [30] Chen, W.B., Feng, W.Q. Yin, J.H. 2020. E˘ects of water content on resilient modulus of a granular material with high fnes content. Construction and Building Materials 236-117542: 1-14, https://doi. org/10.1016/j.conbuildmat.2019.117542. [31] Gabr, A.R., Cameron, D.A. 2012. Properties of recycled concrete aggregate for unbound pavement construction. Journal of Materials in Civil Engi­neering 24(6): 754-764, https://doi.org/10.1061/ (ASCE)MT.1943-5533.0000447. [32] Dong, Q., Huang, B. 2014. Laboratory evaluation on resilient modulus and rate dependencies of RAP used as unbound base material. Journal of Materi­als in Civil Engineering 26(2): 379-383, https://doi. org/10.1061/(ASCE)MT.1943-5533.0000820. [33] da Conceiçao Leite, F., dos Santos Motta, R., Vasconcelos, K.L., Bernucci, L. 2011. Laboratory evaluation of recycled construction and demoli­tion waste for pavements. Construction and Building Materials 25(6): 2972-2979, https://doi. org/10.1016/j.conbuildmat.2010.11.105. [34] Arulrajah, A., Piratheepan, J., Disfani, M.M., Bo, M.W.2013. Resilient moduli response of recycled construction and demolition materials in pave­ment subbase applications. Journal of materials in civil engineering 25(12): 1920-1928, https://doi. org/10.1061/(ASCE)MT.1943-5533.0000766. [35] Rahman, M.A., Arulrajah, A., Piratheepan, J., Bo, M.W., Imteaz, M.A. 2014. Resilient modulus and permanent deformation responses of geogrid-reinforced construction and demolition materials. Journal of Materials in Civil Engineering 26(3): 512-519, https://doi.org/10.1061/(ASCE)MT.1943­5533.0000824. [36] ASTM D422-63. 2009. Standard test method for particle-size analysis of soils. American Society for Testing and Materials, West Conshohocken, USA. [37] ASTM D1241-07. 2007. Standard Specifcation for materials for soil-aggregate subbase, base, and surface courses. American Society for Testing and Materials, West Conshohocken, USA. [38] Ok, B. 2018. Investigation of the behaviour under static and cyclic loading of construction and demolition waste embankment reinforced with geosynthetics. PhD thesis, Cukurova University Institute of Natural and Applied Sciences, Adana, Turkey, pp. 1-414. T. Sarici et al.: The resilient modulus of hybrid construction and demolition wastes reinforced by a geogrid [39] BS EN 933-11. 2009. Tests for geometrical proper­ties of aggregates part 11: classifcation test for the constituents of coarse recycled aggregate. British Standards Institution, London, United Kingdom. [40] ASTM C131–03. 2003. Standard test method for resistance to degradation of small-size coarse aggregate by abrasion and impact in the los angeles machine. American Society for Testing and Materi­als, West Conshohocken, USA. [41] ASTM C127–01. 2001. Standard test method for density, relative density (specifc gravity), and absorption of coarse aggregate. American Society for Testing and Materials, West Conshohocken, USA. [42] ASTM C128–01. 2001. Standard test method for density, relative density (specifc gravity), and absorption of fne aggregate. American Society for Testing and Materials, West Conshohocken, USA. [43] ASTM D4791-10. 2010. Standard test method for fat particles, elongated particles, or fat and elon­gated particles in coarse aggregate. American Soci­ety for Testing and Materials, West Conshohocken, USA. [44] ASTM D1557. 2012. Standard test methods for laboratory compaction characteristics of soil using modifed e˘ort. American Society for Testing and Materials, West Conshohocken, USA. [45] ASTM D1883–99. 1999. Standard test method for CBR (California bearing ratio) of laboratory-compacted soils. American Society for Testing and Materials, West Conshohocken, USA. [46] ASTM D2166. 2006. Standard test method for unconfned compressive strength of cohesive soil. American Society for Testing and Materials, West Conshohocken, USA. [47] Moghaddas-Nejad, F., Small, J.C. 2003. Resilient and permanent characteristics of reinforced granular materials by repeated load triaxial tests. Geotechnical Testing Journal 26(2): 152-166, https://doi.org/10.1520/GTJ11324J. [48] Souci, G. 2009. Laboratory characterization of geogrid-reinforced unbound granular material for use in fexible pavement structures. MSc ~esis Baton Rouge, USA:Louisiana State University, Available online: https://digitalcommons.lsu.edu/ cgi/viewcontent.cgi?referer=https://scholar.google. com.tr/&httpsredir=1&article=1702&context=gra dschool_theses. [49] AASHTO T 307-99. 2000. Determining the resil­ient modulus of soil aggregate materials. Standard Specifcations for Transportation Materials and Methods of Sampling and Testing. ~e American Association of State Highway and Transportation Oycials, Washington D.C., USA. [50] Han, B., Ling, J., Shu, X. et al. 2019. Quantifying the e˘ects of geogrid reinforcement in unbound granular base. Geotextiles and Geomembranes 47(3): 369-376, https://doi.org/10.1016/j. geotexmem.2019.01.009. [51] Hubner, D., Vuong, B., Jameson, G., Sharp, K., Field­ing, B. 2008. Guide to pavement technology part 4a: Granular base and subbase materials, Austroads Ltd., Sydney, Australia. [52] TSH. 2013. Technical Specifcations for Highways. General Directorate of Highways, Ankara, Turkey. [53] Arulrajah, A., Rahman, M.A., Piratheepan, J., Bo, M.W., Imteaz, M.A. 2014. Evaluation of interface shear strength properties of geogrid-reinforced construction and demolition materials using a modifed large-scale direct shear testing apparatus. Journal of Materials in Civil Engineering 26(5): 974-982, https://doi.org/10.1061/(ASCE)MT.1943­5533.0000897. [54] Mehrjardi, G.T., Azizi, A., Haji-Aziz, A., Asdol­lafardi, G. 2020. Evaluating and improving the construction and demolition waste technical properties to use in road construction. Transporta­tion Geotechnics 23-100349: 1-13, https://doi. org/10.1016/j.trgeo.2020.100349. [55] Holtz, R.D., Christopher, B.R., Berg, R.R. 1998. Geosynthetic design and construction guidelines. Federal Highway Administration, Washington, DC, FHWA-HI-98-038, pp. 1-460, Available online: https://ntlrepository.blob.core.windows.net/ lib/21000/21600/21668/PB99130841.pdf. [56] Zornberg, J.G., Prozzi, J.P., Gupta, R. et al. 2008. Validating mechanisms in geosynthetic reinforced pavements. Federal Highway Administration, Texas, USA, FHWA/TX-08/0-4829-1 pp. 1-268, Available online: http://www.utexas.edu/research/ ctr/pdf_reports/0_4829_1.pdf. [57] Edinçliler, A., Yildiz, O. 2018. Seismic behavior of tire waste-sand mixtures for transportation infrastructure in cold regions. Sciences in Cold and Arid Regions, 7(5), 626-631. https://doi. org/10.3724/SP.J.1226.2015.00626. [58] Yildiz, Ö. 2021. Numerical Analysis of Geotechnical Seismic Isolation System for High-Rise Buildings. Naturengs, 2(1), 34-43. https://doi.org/10.46572/ naturengs.872231 14. Acta Geotechnica Slovenica, 2022/2 Ö. Yildiz and Ç. Ceylan: Geotechnical characterization of zeolite-sand and bentonite-sand mixtures GEOTECHNICAL CHARAC­TERIZATION OF ZEOLITE­SAND AND BENTONITE-SAND MIXTURES Özg Yildiz (corresponding author) Malatya Turgut Ozal University, Faculty of Engineering and Natural Sciences Civil Engineering Department Malatya, Turkey E-mail: ozgur.yildiz@ozal.edu.tr GEOTEHNI°NA KARAKTERI­ZACIJA MEŠANIC ZEOLITA IN PESKA TER BENTONITA IN PESKA Çi˛dem Ceylan Malatya Turgut Ozal University, Faculty of Engineering and Natural Sciences Civil Engineering Department Malatya, Turkey https://doi.org/10.18690/actageotechslov.19.2.15-32.2022 zeolite, bentonite, shear strength, correlation, neural networks, prediction ˛is paper presents the characterization of pure bentonite- and zeolite-type clays and of various contents mixed with sand. ˛e engineering properties of zeolites, bentonites and sand, which are commonly found in Malatya, Turkey, were evaluated in terms of their suitability for geotechnical applications. ˛e crystallinity and structure of solid speci­mens of bentonite and zeolite were analysed with X-ray di˙raction. ˛en both soils were mixed with sand in vari­ous proportions and the enhancement of the engineering properties was investigated. ˛e properties of the mixtures, such as specifc gravity, optimum water content, and dry unit weight mixtures, were initially determined. A set of direct shear tests was carried out to determine the shear-strength parameters of the specimens. As a result of exten­sive laboratory tests, linear correlations were observed between the water content and the consistency limits with the bentonite and zeolite contents in the sand mixtures. ˛e highest for among each sample tested was achieved with the addition of 50 % bentonite and zeolite (i.e., BS50 and ZS50) as 44 and 38 kPa, respectively. A literature survey was carried out to reveal the test results of similar studies. In addition, using the test results from these litera­ture studies and the current study, an NN-based predic­tion model was developed. ˛e forecast models developed separately for cohesion and internal friction angle had high correlation coeycients: R2 equal to 0.84 for cohesion and R2 equal to 0.78 for the friction angle. zeolit, bentonit, strižna trdnost, korelacija, nevronske mreže, napovedovanje V prispevku je predstavljena karakterizacija °istih bento­nitnih in zeolitnih glin z razli°nimi vsebnostmi mešanic s peskom. Ocenjene so bile inženirske lastnosti zeolitov, bentonitov in peska, ki jih obi°ajno najdemo v Malatyi v Tur°iji, glede na njihovo primernost za uporabo v geotehniki. Z rentgensko difrakcijo sta bili analizirani kristalini°nost in struktura trdnih vzorcev bentonita in zeolita. Nato sta bili obe zemljini zmešani s peskom v razli°nih razmerjih in raziskano izboljšanje inženirskih lastnosti. Na za°etku so bile dolo°ene lastnosti mešanic, kot so specif°na gravitacija, optimalna vlažnost in suhe prostorninske teže mešanic. Za dolo°itev parametrov strižne trdnosti preizkušancev je bil izveden niz direktnih strižnih preizkusov. Kot rezultat obsežnih laboratorijskih preizkusov so bile opažene linearne korelacije med vlažnostjo in mejami konsistence z vsebnostjo bentonita in zeolita v peš°enih mešanicah. Najvišji kohezijski del strižne trdnosti med posameznimi preizkušanci je bil dosežen z dodatkom 50 % bentonita in zeolita (tj. BS50 in ZS50), in sicer 44 oziroma 38 kPa. Poleg tega je bil z uporabo rezultatov preizkusov študij iz literature in trenutne študije razvit napovedni model osnovan na nevronskih mrežah. Modela napovedi, razvita lo°eno za kohezijo in kot notranjega trenja, imata visoke korela­cijske koefciente, in sicer: R2 enak 0,83 za kohezijo in R2 enak 0,78 za kot notranjega trenja. Ö. Yildiz and Ç. Ceylan: Geotechnical characterization of zeolite-sand and bentonite-sand mixtures 1 INTRODUCTION Soils that can be found freely in nature in di˘erent forms can provide remarkable improvements in terms of engineering and strength properties when combined with di˘erent types of soils or materials. Zeolites are natural and synthetic inorganic aluminosilicates that belong to a large family of open-framework materials consisting of aluminosilicate minerals. One of the most important features of zeolites, which contain a large number of channels and voids, is that they lose the water in these channels at high temperatures without destroying their structure. ~ere are silicon, aluminum, and oxygen in their skeletal structures, and water molecules, alkaline and alkaline-earth cations allow ion exchange in their pores [1]. ~ere are varieties of natural and synthetic zeolites such as clinoptilolite, chabazite, phillipsite and mordenite, which basically have similar molecular structures [2, 3]. Bentonites, on the other hand, are so, porous and easily shaped, open rock, predominantly having a colloidal silica structure and consisting of clay minerals (mainly montmorillonite) with very small crystals formed by chemical weathering or the degradation of volcanic ash, tu˘ and lava rich in aluminum and magnesium. ~ese two soil types, which stand out with their di˘erent structural and mechanical properties, are widely used in engineering applications and are still the subject of detailed experimental studies by researchers. Zeolite, because of its abundance in nature and eco-friendliness, as well as its high potential to increase soil strength, can be a good alternative to a binding material. Due to its high cation-exchange capacity, zeolite can also be used as an adsorbent for the removal of pollutants in wastewater [4]. Besides, it is widely used as a soil-stabilization additive [5, 6, 7, 8, 9]. Yilmaz et al. [10] investigated the e˘ects of zeolite on the mechanical properties of soil under the freeze-thaw e˘ect. Mola-abasi and Shooshpasha [6] performed experiments and numerical modeling studies on the enhancement of the unconfned compressive strength of sand with the inclusion of zeolite. Yukselen and Aksoy [11] proposed zeolite-soil mixtures to be used as embankment- and landfll-liner material. Vogiatzis et al. [12] used Hellenic natural zeolite in mixtures with sand and portland cement. Natural zeolites used instead of sand in mortar mixes decreased the P-wave velocity of sand per unit weight. Mola-abasi et al. [13] investigated the e˘ect of zeolite and cement on the strength of cemented sand specimens. Villalobos et al. [14] stated that zeolites improve the shear strength of the mixtures to which they are added, dependent on their grain size. Bentonites, on the other hand, are defned as clays containing predominantly montmorillonite and have formed as a result of the chemical decomposition of volcanic ash, tu˘, and lava rich in aluminum and magne­sium. Its high swelling capacity is the most important feature that distinguishes bentonites from other clay minerals. Bentonite's properties, such as swelling with water, color, grain size, and moisture absorption ratio, mainly determine its usage areas. ~ey are oen used as an additive material and their physical properties are made use of rather than their chemical properties. Composed of high-swelling montmorillonite, bentonite has been used in various applications such as nuclear-waste dumps, drilling mud, and shear walls due to its water-holding capacity and permeability [15, 16]. To enhance the geotechnical properties of the host material, bentonites are jointly used with fy ash, graphite, basalt, or crushed rock as an additive [17, 18, 19]. ~e hydraulic conductivity of pure bentonite and bentonite-sand mixtures was investigated by considering the di˘erence between the size of both materials [20]. Proia et al. [21] performed experiments with sand-bentonite mixtures of various contents. ~e inclusion of bentonite even at smaller amounts (i.e.,  5 %) reduces the hydraulic conductivity and with the inclusion of higher amounts of bentonites, the mixture becomes more compressible. ~e hydraulic conductivity of sand-bentonite mixtures decreases by four orders of magnitude with the inclusion of 5 % bentonite [22]. Muntohar [23] stated that the existence of bentonite in the soil mixtures infuences the swelling behavior, through a hyperbolic curve model. Alkaya and Esener [24], using various contents of cement and bentonite, revealed that the mixture with 10 % bentonite has the best performance in terms of hydraulic conductivity. Durukan et al. [25] investigated the suction behavior of zeolite-bentonite and sand-bentonite mixtures. In experimental studies where zeolite is used in di˘erent physical forms, it has been observed that as the grain size increases, the suction capacity increases, and zeolite-bentonite mixtures exhibit higher matric suction values than sand-bentonite mixtures. ~e above-mentioned studies demonstrate that both bentonite and zeolite materials have been used in a wide range of applications and investigated in accordance with di˘erent purposes. In many of the studies, zeolite and bentonite were mixed with sand for di˘erent purposes and a summary of the literature survey is given in Table 1. In this study, experimental investiga­tions will be carried out, particularly on zeolite and bentonites, which are two common soil types in the investigation area of the city of Malatya.In the frst place, engineering properties (i.e., grain size, specifc gravity, optimum water content, maximum dry unit 16. Acta Geotechnica Slovenica, 2022/2 Ö. Yildiz and Ç. Ceylan: Geotechnical characterization of zeolite-sand and bentonite-sand mixtures weight, consistency limits and shear strength) were determined. ~e results obtained for both pure and mixed materials were examined to assess the materials’ suitability for geotechnical applications. A literature survey was carried out and the results of similar tests were compiled. Using both literature and current test results, a prediction model was developed. ~e shear parameters of the specimens were estimated using the prediction model. ~e feasibility of NN-based predic­tion models in estimating the shear-strength parameters of multi-component composite materials such as the used soil pairs was demonstrated. 2 GENERAL GEOLOGY ~e city of Malatya is located in eastern Turkey. It has an area of 12,313 km2 (Figure 1). ~e Malatya plain was formed aer the Alpine folding by the fractures and folds during the tectonic movements that emerged at the end of the third geological time and the beginning of the fourth period. It is one of the densest settlements in eastern Anatolia. ~e base rock unit in Malatya and its environs is metamorphites consisting of permo-carboniferous schists and crystallized limestones crop­ping out. In the south of Yeilyurt and Gndbey, there is a conglomerate consisting of red-colored terrestrial conglomerate, sandstone, and mudstone from bottom to top. Inekpinari limestone consists of shallow marine carbonates, the Kapullu formation consists of conglom­erate, sandstone, limestone, and shale alternation, and Haçova formation consisting of tu˘ and andesites exists at the top. On the other hand, in the surrounding Yeilyurt and Gndbey areas, the Yeilyurt group consists of Zorban pebblestone, red-colored conglomer­ate, and sandstones in the form of alluvials from bottom to top, and Yildiz limestone, which consists of reefal limestones. Overlying the Yildiz limestone, the upper Banazi formation with conglomerate, sandstone, and shale alternations emerge. ~is formation is also harmo­niously overlain by Banaz limestones, the Malkuyu formation consisting of marls, and the Gedik formation consisting of reefal limestones. At the bottom of the Lower-Middle Miocene aged terrestrial formations outcropping in the near west, north, and east of Malatya, there is the Akyar formation, which consists of Lower Qal. Alluvium; Qe. Egribuk formation (Sand-gravel-clay); Tqb. Beylereresi formation (Gravelstone-sandstone); Ta. Yazihan group (Limestone, Gravelstone-sandstone-marn); Tsk.- Tsç. Sultansuyu Formation (Limestone-Gravelstone-Sandstone-Marn); Ty-Tyç-Tyb. Yamadagi formation (Gravelstone, clayey limestone, andesite-basalt); Td. Darphane formation (Gravelstone-Limestone); Tp. Petekkaya formation (Sandstone-Marn- Fossiliferous limestone); Tha. Hantarla formation (Gravelstone-sandstone-marn-gypsiums); Th. Haraplar formation (Gravelstone-limestone); Tyk-Tyf. Ye~ilyurt formation (Limestone-gravelstone-sandstone-claystone-marn); Kgk.-Kgs. Hekimhan formation (Limestone); Kgf. Gunduz-bey formation (Gravelstone- claystone-marn); Kgb. Baskil magmatcis (Gabro-diorite); Kg~. Gunes Ophiolites (Serpentine); C-Trmm. Malatya metamorphites (schist-crystalline limestone) Figure 1. Site location and the general geological map of Malatya [26]. Ö. Yildiz and Ç. Ceylan: Geotechnical characterization of zeolite-sand and bentonite-sand mixtures Miocene aged reef limestone and marls. ~e Kuseyin formation, which consists of red conglomerate, sand­stone, mudstone, and gypsum conformably overlies the Akyar formation. ~is Lower Miocene-aged succession is conformably overlain by the Middle Miocene-aged Kilayik, Parçikan, eyhler, Sultansuyu, and Beylereresi formations. ~e general geological map created by the local government oycers is given in Figure 1. ~e zeolites located in the vicinity of Hekimhan, a district of Malatya, are of marine origin and spread over an area of approximately 90 km2. ~e Upper Cretaceous unit is separated into two di˘erent units: the lower zeolite and the upper zeolite unit. ~e lower zeolite unit consists of zeolite with mafc minerals and layers with massive zeolite minerals. Its thickness is at most 15 m and a lateral continuation of 5 km is observed. ~e upper zeolite unit consists of zeolite minerals with sandstone interlayers. Its thickness is at most 38 m and a lateral continuity of 24 km is observed. ~e total geological reserve of the lower and upper zeolite levels is 190 million tons [27]. In addition, it is predicted that there are bentonite reserves at the rate of 50 thousand tons/year in Malatya province Battalgazi, Arapgir, Taskiran and Karahyk districts and localities. It is stated that when the research is expanded to include the surrounding provinces, important reserve areas suitable for the use of di˘erent industries can also be determined. ~e directorates of mineral research and exploration, ayliated with the central government, are actively operating in the region. 3 EXPERIMENTAL STUDY 3.1 Materials and Methods In the experimental studies carried out within the scope of this paper, three di˘erent soils were used, i.e., sand, zeolite and bentonite. ~e selected sand type is widely used in Malatya, especially in the construction industry, and was obtained from the Hekimhan district of Malatya. Zeolite material is freely available in the district of Hekimhan in Malatya. Bentonite is also found freely in nature in the Battalgazi district of Malatya. Both of the materials were supplied in block form; they were grinded and were suitable for our experiments. ~e grain-size-distribution curves of sand, bentonite and zeolite are shown in Figure 2. According to the USCS (Unifed Soil Classifcation System), sand is classifed as SW. ~e bentonite and zeolite are categorized as MH and CH, respectively. Each of the tested materials and mixtures with varies proportions were demonstrated in Figures 3–5. A series of geotechnical laboratory tests were carried out to determine the engineering param­eters of the sand, bentonite, and zeolite, as well as their mixtures with specifed contents. In addition to clean specimens, bentonite and zeolite were mixed in fve di˘erent contents with sand: 10 %, 20 %, 30 %, 40 %, and 50 %. ~e specimens were abbreviated as B, S, Z, BS10, ZS50, etc. ~e letters represent the initials of the components of the mixture. ~e numeral represents the percentage of the additive in the mixture. For instance, BS20 is the abbreviation for the mixture of sand with Figure 2. Grain size distribution of the soils. 18. Acta Geotechnica Slovenica, 2022/2 Ö. Yildiz and Ç. Ceylan: Geotechnical characterization of zeolite-sand and bentonite-sand mixtures Table 1. ~e summary of the recent studies on use of zeolite and bentonite. Case 1 Soil* B/S B/S Content (%) 10/90 20/80 PI (%) 13.7 59.3 wopt (%) ^ (kN/m3) C (kPa) 3 10 ˘ (°) 28.7 19.6 Reference [33] 2 3 B/S B/S B/S B/S B/S B/S Z/S 30/70 40/60 50/50 70/30 10/90 20/80 25/75 98.9 157.6 201.7 312.4 18.6 19 10.14 16.1 15.63 19.33 6 7 5 6 8.7 5.6 3.8 3.8 [34] [35] Z/S 50/50 18.26 15.93 4 Z/S B/S 75/25 50/50 27.03 22.5 13.89 15.35 [36] B/S 60/40 19 15.96 B/S 70/30 16 16.3 B/S 80/20 15.1 16.77 B/S 90/10 14.5 16.39 B/S 50/50 22.5 15.54 B/S 60/40 20.5 15.64 B/S 70/30 18 16.23 B/S 80/20 18.4 16.68 B/S 90/10 17.2 16.08 5 B/S 15/85 52 17 16.6 [37] B/S 25/75 70 15 17,2 6 B/S 3/97 10 19.35 6.43 47 [22] B/S 5/95 10.5 19.1 21.47 37 B/S 7/93 11.2 18.68 24.11 35 B/S 9/91 12 18.56 24.9 33 7 B/S 20/80 15.28 1727 16.4 24.9 [38] 8 B/S 15/85 115 15 17.3 [39] B/S 25/75 231 15.8 17.2 B/S 50/50 333 20 15.2 9 B/S 70/30 59 27 15.1 [40] B/S 60/40 46 22 15.9 B/S 50/50 30 18 16.6 10 B/S 5/95 19.4 15.79 [41] B/S 10/90 17.6 16.08 B/S 20/80 17 16.47 B/S 30/70 14.6 16.87 B/S 50/50 17.5 16.28 11 Z/S 25/75 16.5 17.5 30.6 37.3 [42] Z/S 50/50 20 16.8 32.5 35.8 Z/S 75/25 22.5 15.7 31.2 31.7 12 Z/S 5/95 3.85 9 20.08 31.23 [43] Z/S 10/90 3.848 10.2 19.5 31.48 Z/S 15/85 3.328 11.5 18.7 32.55 Z/S 20/80 3.08 12.3 18.3 33.34 Z/S 25/75 2.92 13 18 33.18 Z/S 30/70 3.84 13.8 17.93 33.27 Z/S 35/65 4.16 15.3 17.25 33.29 * Z, B and S denote the zeolite, bentonite and sand, respectively. Ö. Yildiz and Ç. Ceylan: Geotechnical characterization of zeolite-sand and bentonite-sand mixtures 20 % bentonite. Grain-size-distribution analyses, compaction tests, consistency limit tests, permeability tests, and direct shear tests were performed in accor­dance with ASTM D422-63, ASTM D698, ASTM D4318, ASTM D2434-94, and ASTM D3080-98, respectively [28, 29, 30, 31, 32]. ~e clean sand specimen used in the experiments was le to dry at room temperature in a laboratory environment. ~e dried specimens were sieved with # 4 (4.75 mm) and # 200 (0.075 mm) sieves, and the material remaining between the number # 4 and # 200 sieves were used in the experiments. Bentonite and zeolite specimens were taken from a depth of 1.5 to 2 m from the surface and le to dry at room temperature. ~e dried specimens were grinded in a ball mill and sieved through sieve # 200. In the experiments, materi­als fner than 0.075 mm were used. ~e materials were prepared by dry mixing the bentonite and zeolite with sand separately at the specifed mixing ratios (i.e., 10 %, 20 %, 30 %, 40 %, and 50 %). While the samples for the consistency limit test were kept in a desiccator overnight, the samples soaked in the standard proctor test were subjected to the test by keeping them in sealed bags for at least 3 hours. ~e specimens that were moistured and compressed at an optimum water content were used in direct shear tests. Pure water was used for wetting speci­mens by spraying to form a homogeneous mixture. Figure 3. Pure materials used in the experimental study. Figure 4. BS specimens with various bentonite contents. 20. Acta Geotechnica Slovenica, 2022/2 Ö. Yildiz and Ç. Ceylan: Geotechnical characterization of zeolite-sand and bentonite-sand mixtures Figure 5. ZS specimens with various zeolite contents. 4 DISCUSSION AND RESULTS 4.1 Laboratory tests Prior to the geotechnical laboratory tests, mineralogy and microscopic analyses of the zeolite were carried out. ~e identifcation of the zeolites using X-ray techniques is diycult because of the di˘erent cell dimensions and the di˘erences in the relative intensities of the bands [11]. As can be seen in the XRD pattern in Figure 6, the zeolite has a high concentration of quartz and lower calcite and clinoptilolite content. controlled XRD analyzes were performed in InnUniversity laboratories (IBTAM). ~e basis of the work was to detect di˘erent crystal structures or the parameters in crystalline materials based on the refection (refraction) of the x-ray. ~e beam is refected (i.e. or refracted) on the sample and the beam detected with the help of a detector is transferred to the graph with the 2 value corresponding to the refection intensity using soware. To determine the mineralogical compositions of the raw materials used the materials were prepared by passing Figure 6. X-ray di˘raction spectra of the (a) zeolite, (b) bentonite. Ö. Yildiz and Ç. Ceylan: Geotechnical characterization of zeolite-sand and bentonite-sand mixtures through a 150-µm sieve. ~e X-rays were detected with a RigakuRadB-DMAX II computer-controlled X-ray di˘ractometer using Cu–K radiation. Measurements were scanned between 2 = 3° to 80° degrees and at a constant speed of 3°/min. ~e analyses were performed according to the ASTM D5758 standard [44]. ~e di˘ractogram of natural zeolite shows the intensity at a 2 angle of 26° with a peak of 3100 counts correspond­ing to the presence of quartz (SiO2), which is a very common and important mineral. ~e bentonite, on the other hand, includes montmorillonite at a 2 angle of 22° with a peak of 235 counts. ~e second-most intense mineral was found to be feldspar in bentonite. ~e specifc gravities of the sand, bentonite and zeolite were calculated as 2.69, 2.46 and 2.38, respectively. Accord­ingly, the increasing content of both bentonite and zeolite leads to a decrease in the specifc gravity of the mixtures (Figure 7). As a host material, when the sand is mixed with bentonite or zeolite, a soil mass is formed in which sand particles make up the skeleton structure and additive particles occupy the voids in the matrix [41]. ~e size, distribution and compressibility of these voids are mainly dependent on the size, shape and propor­tions of sand particles in the mixture [36, 45]. Also, the mineralogy, content, compaction energy applied and moisture content are the infuencing parameters for the mechanical characteristics of the compacted specimens [36, 41, 46, 47, 48, 49]. A set of modifed proctor compaction tests was carried out and the optimum water content for the maximum compaction and unit weight was obtained for the specimens. As can be seen, the amount of water required to obtain the maximum unit weight is increasing with the increas­ing bentonite and zeolite content. ~e maximum unit weight of the BS10 specimen is 2.1 g/cm3 for 12.3 % of water inclusion. ~e BS50 specimen including 50 % of bentonite in the mixture reaches the maximum unit weight as 1.67 g/cm3 with 21 % of water content (Figure 8). In bentonite-sand mixtures, the values reached regarding the optimum water content are higher than that of the zeolite-sand mixtures (Figure 9). ~is is most likely because the included bentonite specimens have a relatively large surface area that causes a higher amount of water absorption than the included zeolite specimens. Since the bentonite particles are fner than the zeolite particles, the pores between the sand grains are reduced more easily in the BS specimens. ~erefore, the optimum moisture content of the bentonite sand mixtures is higher than that of the zeolite-sand mixtures for the same additive content. ~e highest unit weight can be achieved with less water content for the ZS specimens. For example, the unit weight of the BS50 specimen is observed to be 14 % lower than that of ZS50 (i.e., 1.90 to 1.67 g/cm3). ~e amount of water for the BS50 and ZS50 specimens is 21 % and 15 %, respectively. Even the compressibility of the clay-coarse-soil mixture is assumed to be dependent on the complex physicochemical interactions of the clay particles and the contribution of the mechanical properties of coarse soil (Bolt, 1956), a very clear pattern observed by the compaction curves [46]. Figure 10a shows the variation of the optimum water content needed to achieve the maximum unit weight for each content of bentonite and zeolite in the sand. ~e variations in the optimum water content due to bentonite and zeolite addition to the sand can be clearly observed. As the amount of inclusion increases, the optimum water content increases. Both the compaction curves of the BS and ZS specimens were following a defnite pattern. As can be seen from the regression curves, the optimum water content for the maximum 22. Acta Geotechnica Slovenica, 2022/2 Ö. Yildiz and Ç. Ceylan: Geotechnical characterization of zeolite-sand and bentonite-sand mixtures Figure 8. Compaction curves of the bentonite-sand mixtures. Figure 9. Compaction curves of the zeolite-sand mixtures. compacted unit weight in correlation with bentonite/ zeolite content in the mixtures. ~e amount of water required to bring the mixtures to the maximum unit weight is much less for the ZS specimens than for the BS specimens (Figure 10b). ~is is necessarily related to the di˘erence between the gradational parameters and the compactional characteristics of both materials. It was observed that the water-holding capacity of bentonite is higher than zeolite for the same mixing ratios. In other words, the water-adsorption capacity of bentonite is higher than that of zeolite. ~e maximum unit weight of specimens decreases with the increas­ing water content. It should also be remembered that bentonite is used in engineering applications as a dispersive material. ~e high correlation coeycients between the content of both bentonite and zeolite in the mixture and the optimum water content clearly show how much the compaction behavior is suppressed by the additive content in the mixture. Depending on the types of soil used, linear relationships between the optimum water content, the maximum unit weight and the applied compaction energy were also developed with similar studies [24, 34, 50, 51, 52]. ~e plasticity characteristic of the mixtures depends on the content and type of mineral in the additive [53]. In some studies, it has been suggested that at low clay contents, the mixture exhibits predominantly granular properties, while higher ratios a gradual transition to mechanical behavior of the plastic clay occurs [21]. However, Bowles [54] stated that the addition of 2 % clay to sand is the initial value for transforming the Ö. Yildiz and Ç. Ceylan: Geotechnical characterization of zeolite-sand and bentonite-sand mixtures mixture from a sandy state to a clayey state. It is, therefore, the amount of zeolite and bentonite in the mixture was set at higher contents in this study. ~e variation of the consistency limits with additive content is presented in Figure 11. ~e consistency parameters of the mixtures increase with increasing both bentonite and zeolite content. ~e zeolite at a higher content of 20 % displayed an increase in the plasticity of the mixtures. ~e ine˘ectiveness of the zeolite addition at less than 6–10 % on the plasticity is attributed to chemical properties such as the sodium absorption ratio (SAR) and exchangeable sodium percentages (ESP) [55]. ~e bentonite, on the other hand, even with smaller contents, has led to an increase in plasticity. ~e montmorillonite included in the bentonite has a key role in its plastic behavior. As can be seen from the high correlation coeycient, the variation between the content and the liquid limit is almost linear 24. Acta Geotechnica Slovenica, 2022/2 Ö. Yildiz and Ç. Ceylan: Geotechnical characterization of zeolite-sand and bentonite-sand mixtures (Figure 11a). ~e regression lines show that the inclu­sion of additives has a lower e˘ect on the plastic limit than on the liquid limit (Figure 11b). Bowles [54] also stated that the increasing bentonite inclusion into sand leads to a linear increase in the liquid limit, but has a limited e˘ect on the plastic limit. It is clear that zeolite is more e˘ective than bentonite on the plasticity index at mixing ratios greater than 20 % (Figure 11c). ~e plasticity characteristics of the sand-bentonite mixtures are dependent on the clay content and the clay-mineral type [56]. ~e granulometry and mechanical charac­teristics are also found to be factors infuencing the plasticity [57]. It is assumed that the plasticity e˘ect induced by the addition of zeolite and bentonite, even with the same content, is not at the same level. ~e Casagrande plasticity chart built with the consistency limits of the specimens is shown in Figure 12. ~e specimens were obtained by artifcially mixing soils with di˘erent physical and mechanical properties. ~us, the advantages of both materials can be combined by trying combinations with di˘erent contents [21]. It is important to examine the shear strength of the BS and ZS specimens formed by mixing at di˘erent ratios. ~e shear-strength parameters of the specimens are signif­cant, especially for a stability analysis. In the case of their use as liners or backfll materials, the shear strength of the zeolites and bentonites was investigated for both the drained and undrained cases. In this study the cohe­sion and internal friction angle of the specimens were determined through undrained direct shear tests. ~e compacted BS and ZS specimens were sheared immedi­ately aer compaction. ~e normal stresses applied as 28, 56 and 111 kPa. A strain rate of 0.5 mm/min was used for all tests and the time required for shearing the specimens to failure was about 10 to 15 min. For each content of zeolite and bentonite, the test results demonstrated that increasing the applied normal stress leads to an increase in the shear stress. It was also observed that the measured maximum shear stress decreases with increasing additive content. ~e mixture with a 50 % inclusion of bentonite and zeolite (i.e., BS50 and ZS50) had the best performance as 92 and 87 kPa Ö. Yildiz and Ç. Ceylan: Geotechnical characterization of zeolite-sand and bentonite-sand mixtures under 111 kPa normal stress, respectively (Figure 13). ~e specimens having smaller contents of additives display a hardening behavior during shearing, which is more visible for the ZS specimens. ~e variation of the maximum shear stress with the normal stress is shown in Figure 14. Specimens containing both bentonite and zeolite under lower normal stresses show a higher shearing response than the clean sand. However, when the applied normal stress increases from 28 kPa to 111 kPa, the response of the clean sand and the mixtures to shearing becomes closer. Under higher normal stresses, only specimens with a greater content of bentonite and zeolite (i.e., BS40, BS50 and ZS50) have higher shear stresses than clean sand. ~is shows that the shear behavior of the mixtures is sensitive to the applied normal stress, so it makes sense to interpret the shear behavior in two parts. In contrast to some litera­ture studies, the specimens appear to exhibit sand-like behavior at high bentonite and zeolite mixing ratios in terms of shear behavior. ~e variation of the engineering properties of the speci­mens with multi-component soils is directly a˘ected by the proportional distribution of the soil types in the mixture. As previously mentioned, one of the main motivations of this study is to combine di˘erent soil types and to take advantage of the better engineering properties of each of the components. ~erefore, triple graphical representations of direct shear test results are given in Figure 15. While the shear stress was measured under 111 kPa normal stress with clean sand and pure bentonite was 86 and 73 kPa, respectively, this value increased to 92 kPa with the BS50 specimen. ~is situation occurred similarly to the ZS50 sample (i.e., 87 kPa) (Figure 15b). ~e shear strength of the speci­mens consists of two components: the cohesion and the internal friction angle. When bentonite and zeolite as cohesive materials were combined with sand, a decrease in the cohesion values was measured, as expected. Cohesion values of 67 and 69 kPa, measured for pure bentonite and pure zeolite, were decreased to 44 kPa and (a) Bentonite and BS specimens, (b) Zeolite and ZS specimens. 26. Acta Geotechnica Slovenica, 2022/2 Ö. Yildiz and Ç. Ceylan: Geotechnical characterization of zeolite-sand and bentonite-sand mixtures 38 kPa for the BS50 and ZS50 specimens, respectively (Figure 15c-d). However, the internal friction angles of the pure specimens, which were 3° and 2°, increased to 23° and 33°, respectively. When bentonite is mixed with sand, due to its very small particle size it occupies the pore space present between the individual sand grains which is also valid for zeolite [53]. ~e optimum amount of material replacement by zeolite or bentonite for the highest improvement in the shear-strength parameters was also investigated by di˘erent researchers [4, 58, 59, 60]. It is essential to determine the optimum content of the materials, which will meet the design target, with both strength values and other engineering properties. ~e basic engineering properties, compac­tion, consistency limits and direct shear test results of all specimens are collectively given in Table 3. 5 PREDICTION MODEL So-computing methods, which are used in the analysis of multivariate and multi-parameter numerical problems that are diycult to interpret with analytical models, are widely used in almost every feld. ~ese methods have Ö. Yildiz and Ç. Ceylan: Geotechnical characterization of zeolite-sand and bentonite-sand mixtures Table 2. Summary of the geotechnical experiment results. Specimen Gs LL (%) PL (%) PI (%) wopt (kPa) dmax (g/cm3) C (kPa) ˘ ( ) USCS S 2.69 NP NP NP - 1.34 6 36 SW Z 2.38 81.5 26.5 55.0 20.4 1.49 69 2 CH B 2.46 62.3 41.0 21.3 32 1.22 67 3 MH BS10 2.65 21.1 16.5 4.5 12.3 2.10 16 30 ML BS20 2.63 30.1 21.2 8.9 13.5 1.98 24 26 CL BS30 2.51 39.9 23.6 16.3 17.1 1.87 30 25 CL BS40 2.46 40.2 22.5 17.7 19.1 1.78 39 24 CL BS50 2.44 44.9 28.8 16.1 21.2 1.67 44 23 ML ZS10 2.69 NP NP NP 10.5 2.12 12 33 NP ZS20 2.69 NP NP NP 11 2.05 15 30 NP ZS30 2.68 23.5 16.5 7.0 14.4 1.92 23 28 CL ZS40 2.65 29.1 15.5 13.6 14.6 1.93 29 26 CL ZS50 2.63 36.1 15.5 20.6 15.0 1.90 38 24 CL also been widely used by researchers in geotechnical engineering [61, 62, 63, 64, 65, 66]. Also, a comprehensive literature survey was carried out on the use of neural networks in geotechnics [67]. Prediction models were developed with the experimental results obtained from the literature review and this study. ~ese models consist of an input layer with 6 parameters, a hidden layer and an output layer with target parameters. ~e input parameters are the soil types and their ratios in the mixture, the plasticity index, the optimum water content and the unit weight. ~e target parameters are set as the shear-strength parameters: the cohesion and the internal friction angle. ~e fowchart of the developed model is presented in Figure 16. Two sets of predictions were made separately with the prediction model developed for both the cohesion and the internal friction angle. As a result of a trial-and-error process, 10 neurons were identifed in the hidden layer. A feed-forward error back propagation model is developed using the Levenberg Marquardt algorithm. ~e architecture of the model is given in Figure 17. Laboratory test results are displayed as target values on the x-axis, and numerical analysis results are displayed on the y-axis. A performance evaluation of the model was made using MSE and square of correlation coef­fcient (R2). ~e linear output indicates the success of the predictive model. In fact, the prediction models work separately on each data set randomly divided for training, validation and testing, and the correlation coeycients and MSE are calculated individually for each stage. However, the overall performance is represented by combining each of the three cases in one graph. ~e regression curves of the predictions for both target parameters, i.e., cohesion and frictional angle, were presented separately in Figure 18. Correlation coey­cients for the measured and estimated cohesion and fric­tion angle were obtained as 0.84 and 0.78, respectively. ~ese success performances, which were developed with a limited number of data sets, showed a reasonable esti­mation of success. Although it was obtained from di˘er- 28. Acta Geotechnica Slovenica, 2022/2 Ö. Yildiz and Ç. Ceylan: Geotechnical characterization of zeolite-sand and bentonite-sand mixtures ent studies and consisted for a limited number of data, an acceptable success performance was obtained with the studied dataset. It is convenient in that it shows that the strength parameters can be estimated practically by so-computing methods using the principle engineering properties. ~e statistical data of the predictions were summarised in Table 3. Table 3. Statistical data of the predicted parameters. Cohesion Angle of friction MSE R MSE R Training phase 10.8881 0.9050 77.4662 0.8303 Validation phase 48.7666 0.7643 33.7122 0.8802 Test phase 45.0875 0.7929 18.0818 0.2991 6 CONCLUSION ~is study was carried out to determine the geotechni-cal properties of pure zeolite and bentonite and their mixtures with sand, which are two common local soil types in the investigated area. In this context, the geotechnical properties and mineralogical properties of the materials were determined. In order to examine the improvements in the shear-strength parameters, direct shear tests were carried out on the combination of locally supplied sand with di˘erent contents. A NN-based model has been developed for the predic­tion of the shear-strength parameters of mixtures with existing geotechnical properties, with both results of literature studies and the current study. ~e main results are drawn in this study as follows: Ö. Yildiz and Ç. Ceylan: Geotechnical characterization of zeolite-sand and bentonite-sand mixtures  A signifcant increase was observed in the shear­-strength parameters of the sand specimens with a mixture of zeolite and bentonite. ~e shear-strength parameters of the mixtures increase proportionally with increasing zeolite and bentonite content.  ~e improvement in the shear-strength parameters with the addition of zeolite and bentonite is much more pronounced under low normal stresses. As the applied normal stress increases, the shear strength of both mixtures and the pure sand draw closer to each other.  ~e maximum cohesion and friction angle measured for the BS50 and BS10 specimens were 44 kPa and 30°, respectively. ~ose parameters were measured by ZS50 and ZS10 specimens as 38 kPa and 24°, which indicates a remarkable di˘erence in favor of the BS specimens in terms of the shear-strength parameters.  Among the tested specimens it was observed that the BS40, BS50 and ZS50 specimens exhibited the highest strength values. ~ese specimens signifcantly increased the plasticity properties of the clean sand in mixtures.  With the compiled database, an acceptable accu­racy was obtained regarding the estimation of the cohesion and the friction angle. ~e correlation coeycient R2 = 0.84 was obtained for cohesion and R2 = 0.78 was obtained for the internal friction angle, which shows the eyciency of the prediction models developed for multi-component soil specimens. REFERENCES [1] Breck, D.W. (1974). Zeolite Molecular Sieves, Struc­ture, Chemistry and Use, John Wiley & Sons, New York. [2] Ören, A. H., & Özdamar, T. (2013). Hydrau­lic conductivity of compacted zeolites. Waste Manag. Res. 31(6):634-40. DOI: 10.1177/0734242X13479434. [3] Dyer, A. (1988). An introduction to zeolite molecu­lar sieves. Chichester; New York: J. Wiley. [4] Norouznejad, G., Shooshpasha, I., Mirhosseini, S. M., Afzalirad, M., & Afzalirad, M. (2021). Infuence of Zeolite on the Compaction Characteristics and Shear Strength Parameters of Cemented Sand. Sains Malaysiana, 50(11), 3181–3191. https://doi. org/10.17576/jsm-2021-5011-03 [5] Savas, H. (2016). Consolidation and swell charac­teristics of dispersive soils stabilized with lime and natural zeolite. Sci Eng Compos Mater 26:589–598. https:// doi. org/ 10. 1515/ secm- 2014- 0202. [6] Mola-Abasi, H., & Shooshpasha, I. (2017). Poly­nomial models controlling strength of zeolite-cement-sand mixtures. Scientia Iranica, 24(2), 526–536. [7] Eyo, E. U., Ng’ambi, S., & Abbey, S. J. (2020). Performance of clay stabilized by cementitious materials and inclusion of zeolite/alkaline metals-based additive. Transportation Geotechnics, 23(February), 100330. https://doi.org/10.1016/j. trgeo.2020.100330 [8] Shahriarkian, M., Kabiri, S., & Bayat, M. (2021). Utilization of zeolite to improve the behavior of cement-stabilized soil. Int J Geosynth Ground Eng 7(2). https:// doi. org/ 10. 1007/ s40891- 021- 00284-9 [9] Chenarboni, H.A., Lajevardi, S. H., MolaAbasi, H., & Zeighami, E. (2021). ~e e˘ect of zeolite and cement stabilization on the mechanical behavior of expansive soils. Constr Build Mater 272:121630. https:// doi. org/ 10. 1016/j. conbuildmat. 2020. 121630 [10] Yilmaz, F., Sado lu, E., & Kamilo lu, H. A. (2022). Evaluation of the e˘ect of waste zeolite on the strength and micro-macrostructure of a high plas­ticity clayey soil stabilized with lime-waste zeolite mixtures subjected to freezing–thawing cycles. Arabian Journal of Geosciences, 15(6). https://doi. org/10.1007/s12517-022-09767-z [11] Yukselen-Aksoy, Y. (2010). Characterization of two natural zeolites for geotechnical and geoenviron-mental applications. Applied Clay Science, 50(1), 130–136. https://doi.org/10.1016/j.clay.2010.07.015 [12] Vogiatzis, D., Kantiranis, N., Filippidis, A., Tzamos, E., & Sikalidis, C. (2012). Hellenic natural zeolite as a replacement of sand in mortar: Mineralogy monitoring and evaluation of its infuence on mechanical properties. Geosciences (Switzerland), 2(4), 298–307. [13] Mola-Abasi, H., Kordtabar, B., & Kordnaeij, A. (2016). Parameters controlling strength of zeolite– cement–sand mixture. International Journal of Geotechnical Engineering, 11(1), 72–79. https:// doi.org/10.1080/19386362.2016.1186412 [14] Villalobos, F. A., Leiva, E. A., Jerez, Ó., & Poblete, M. E.(2018). Experimental study of the fne particles e˘ect on the shear strength of tu˘ zeolites. Revista de La Construccion, 17(1), 23–37. https://doi. org/10.7764/RDLC.17.1.23 [15] Murray, H. H. (2006). Bentonite Applications. Developments in Clay Science, 2(C), 111–130. [16] Sun, Y., Fan, W., Zheng, H., Zhang, Y., Li, F., & Chen, W., (2015) Evaluation of Dewatering Performance and Fractal Characteristics of Alum Sludge, PLoS ONE, Vol. 10(6),1-16. 30. Acta Geotechnica Slovenica, 2022/2 Ö. Yildiz and Ç. Ceylan: Geotechnical characterization of zeolite-sand and bentonite-sand mixtures [17] Komine, H., & Ogata, N. (1999). Experimental study on swelling characteristics of sand-bentonite mixture for nuclear waste disposal. Soils Found. 39 (2): 83–97. Komine, H. 2004.“Simplifed evaluation on hydraulic conductivities of sand–bentonite mixture backfll.” Appl. Clay Sci. 26 (1–4): 13–19. https://doi.org/10.1016/j.clay.2003.09.006. [18] Jobmann, M., & Buntebarth, G. (2009). Infuence of graphite and quartz addition on the thermo-physical properties of bentonite for sealing heat-generating radioactive waste. Applied Clay Science, 44(3–4), 206–210. https://doi.org/10.1016/j. clay.2009.01.016 [19] Chen, T., Sedighi, M., Jivkov, A. P., & Seetharam, S. C.(2021). A model for hydraulic conductivity of compacted bentonite - Inclusion of microstructure e˘ects under confned wetting. Geotechnique, 71(12), 1071–1084. https://doi.org/10.1680/ jgeot.19.P.088 [20] Sivapullaiah, P. V., Sridharan, A., & Stalin, V. K. (2000). Hydraulic conductivity of bentonite-sand mixtures. Canadian Geotechnical Journal, 37(2), 406–413. https://doi.org/10.1139/t99-120 [21] Proia, R., Croce, P., & Modoni, G. (2016). Experi­mental Investigation of Compacted Sand-bentonite Mixtures. Procedia Engineering, 158, 51–56. [22] Chalermyanont, T., & S. Arrykul. (2005). Compacted sand-bentonite mixtures for hydraulic containment liners. Songklanakarin J. Sci. Technol. 27 (2): 313–323. [23] Muntohar, A.S. (1999). Swelling and compress­ibility characteristics of compacted sand-bentonite mixtures. Geotechnical Engineering, 30(1), 1–12. [24] Alkaya, D., & Bari Esener, A. (2011). Usability of sand-bentonite-cement mixture in the construc­tion of unpermeable layer. Scientifc Research and Essays, 6(21), 4492–4503. https://doi.org/10.5897/ sre10.1189 [25] Durukan, S., Pulat, H. F., & Yukselen-Aksoy, Y. (2014). Suction characteristics of compacted zeolite-bentonite and sand-bentonite mixtures. Waste Management and Research, 32(2), 149–156. https://doi.org/10.1177/0734242X13518958 [26] ÇB, (2011). Malatya Province Environment Status Report. Ministry of Environment and Urbaniza­tion, Malatya. [27] Dussan, A., Caldern, J. A., & Quiroz, H.P. (2020). Zeolites derived from natural minerals: Solid and volcanic ash., pp. 148-149. DOI:10.1016/j. mattod.2020.03.006 [28] ASTM D422-63, 2003. Standard Test Method for Particle-Size Analysis of Soils, In Annual Book of ASTM Standards, Volume 04.08, Philadelphia, PA, pp. 93-99. [29] ASTM D-698,“Standard Test Methods for Labora­tory Compaction Characteristics of Soil Using Standard E˘ort,” American Society for Testing and Materials, USA [30] ASTM D4318-17e1. Standard Test Methods for Liquid Limit, Plastic Limit, and Plasticity Index of Soils; ASTM International: West Conshohocken, PA, USA, 2017. [31] ASTM D 2434-94. Standard test method for permeability of granular soils (constant head). Annual Book of ASTM Standards, American Soci­ety For Testing and Materials, West Conshohocken, PA (2000). [32] ASTM D 3080-98, 2003. Standard Test Method for Direct Shear of Soils Under Consolidated Drained Conditions, In: Annual Book of ASTM Standards, Volume 04.08, Philadelphia, PA, pp. 417-422. [33] Tiwari, B., & Marui, H. (2003). Estimation of resid­ual shear strength for bentonite-kaolin-Toyoura sand mixture. Journal of the Japan Landslide Society, 40(2), 124–133. https://doi.org/10.3313/ jls.40.124 [34] Ören, A. H., Kaya, A., & Kayalar, A. . (2011). Hydraulic conductivity of zeolite-bentonite mixtures in comparison with sand-bentonite mixtures. Canadian Geotechnical Journal, 48(9), 1343–1353. https://doi.org/10.1139/t11-042 [35] Shang, H. (2015). Geotechnical laboratory charac­terization of sand- zeolite mixtures, MSc. ~esis, University of Louisville. [36] Vadlamudi, S., & Mishra, A. K. (2018). Consolida­tion characteristics of sand–bentonite mixtures and the infuence of sand particle size. Journal of Hazardous, Toxic, and Radioactive Waste, 22(4), 06018001. [37] Ghadr, S., & Assadi-Langroudi, A. (2018). Structure-based hydro-mechanical properties of sand-bentonite composites. Engineering Geology, 235, 53-63. doi.org/10.1016/j.enggeo.2018.02.002 [38] Mukherjee, K., & Mishra, A. K. (2019). Hydraulic and mechanical characteristics of compacted sand–bentonite: Tyre chips mix for its landfll application. Environment, Development and Sustainability, 21(3), 1411-1428. [39] Chen, Y., & Meehan, C. L. (2011). Undrained strength characteristics of compacted bentonite/ sand mixtures. In Geo-Frontiers 2011: Advances in Geotechnical Engineering (pp. 2699-2708). [40] Fattah, M. Y., Salim, N. M., & Irshayyid, E. J. (2022). Infuence of soil suction on swelling pressure of bentonite-sand mixtures. European Journal of Environmental and Civil Engineering, 26(7), 2554­2568. DOI: 10.1080/19648189.2017.1320236 [41] Komine, H. (2004). Simplifed evaluation on Ö. Yildiz and Ç. Ceylan: Geotechnical characterization of zeolite-sand and bentonite-sand mixtures hydraulic conductivities of sand–bentonite mixture backfll. Applied clay science, 26(1-4), 13-19. [42] Kwan, P.S., Shahrokhi, R., Park, J., Kim,H., (2019). Zeolite mixtures as adsorptive fll material with sustainable bearing capacity. Waste Management and Environment IX, WIT Transactions on Ecol­ogy and the Environment, 231, 91-100. [43] Aabi, S., Fathi, S., & Aminfar, M. H. (2020). ~e E˘ect of Zeolite on Sandy-Silt Soil Mechanical Properties. International Journal of Geotechnical and Geological Engineering, 14(10), 269-278. [44] ASTM D5758-01 Test Method for Determina­tion of Relative Crystallinity of Zeolite ZSM-5 by X-Ray Di˘raction [45] Elsbury, B. R., D. E. Daniel, G. A. Sraders, & D. C. Anderson. (1990). Lessons learned from compacted clay liner.” J. Geotech. Eng. 116 (11): 1641–1660. https://doi.org/10.1061/(ASCE)0733­9410(1990)(1641) [46] Bolt, G. H. (1956). Physico-chemical analysis of the compressibility of pure clays.” Géotechnique 6 (2): 86–93. https://doi.org/10.1680/geot.1956.6.2.86. [47] Sridharan, A., & G. V. Rao. (1973). Mechanisms controlling volume change of saturated clays and the role of the e˘ective stress concept. Geotech­nique 23 (3): 359–382. [48] Mitchell, J. K., & Soga, K. (2005). Fundamentals of soil behavior. 3rd ed. Hoboken, NJ: Wiley. [49] Dutta, J., & A. K. Mishra. (2016). Consolidation behaviour of bentonites in the presence of salt solutions. Appl. Clay Sci. 120: 61–69. https://doi. org/10.1016/j.clay.2015.12.001. [50] Boutwell, G. (1961). E˘ects of variation of fll construction on the material properties and the subsequent fll performance, Independent study Rep., School of Civ. Engrg., Georgia Institute of Technology, Atlanta, Georgia, USA. [51] Blotez, L., Benson, C. & Boutwell, G. (1998). Esti­mating Optimum Water Content and Maximum Dry Unit Weight for Compacted Clays, J Geotech. Engrg., ACSE, 124(9), 907- 912. [52] Jesmani, M., Manesh, A. N., & Hoseini, S. M. R. (2008). Optimum water content and maximum dry unit weight of clayey gravels at di˘erent compac­tive e˘orts. Electronic Journal of Geotechnical Engineering, 13 L(June). [53] Srikanth, V., & Mishra, A. K. (2016). A Laboratory Study on the Geotechnical Characteristics of Sand–Bentonite Mixtures and the Role of Particle Size of Sand. International Journal of Geosynthet­ics and Ground Engineering, 2(1), 1–10. https:// doi.org/10.1007/s40891-015-0043-1 [54] Bowles, J.E. (1998). Engineering Properties of Soil and ~eir Measurement, Mcgraw-Hill Book Company, Singapore. [55] Turkoz, M., & Vural, P. (2013). ~e e˘ects of cement and natural zeolite additives on problematic clay soils. Science and Engineering of Composite Mate­rials, 20(4), 395–405. [56] Sivapullaiah, P.V., & Sridharan, A. (1985). Liquid limit of soil mixtures. Geotech Test J 8(3):111–116. [57] Harianto, T. (2022). Performance of Subbase Layer with Geogrid Reinforcement and Zeolite-Waterglass Stabilization. Civil Engineering Journal (Iran), 8(2), 251–262. https://doi.org/10.28991/ CEJ-2022-08-02-05 [58] Sinha, A. N. (1998). Shear Strength of Bentonite-Kaolinite Mix in GCLs. pp, 889–896. [59] Xiang, Gs., Ye, Wm. & Jalal, F.E. (2021). Shear strength of bentonite–sand mixture saturated with saline solution. Environ Earth Sci 80, 770. [60] Cabalar, A. F., & Demir, S. (2022). Geotechnical properties of a bentonite treated with waste glass grains. Arabian Journal of Geosciences, 15(9). https://doi.org/10.1007/s12517-022-10169-4 [61] Goh, A. T. C. (1994). Nonlinear modelling in geotechnical engineering using neural networks. Australian Civil Engineering Transactions, CE36(4), 293-297. [62] Goh, A. T. C. (1995). Empirical design in geotech­nics using neural networks. Geotechnique, 45(4), 709-714. [63] Cabalar, A. F., Cevik, A., & Gokceoglu, C. (2012). Some applications of adaptive neuro-fuzzy infer­ence system (ANFIS) in geotechnical engineering. Comput Geotech 40:14–33 [64] Yildiz, Ö., & Berilgen, M. M. (2020). Artifcial Neural Network Model to Predict Anchored-Pile-Wall Displacements on Istanbul Greywackes, Teknik Dergi, 31 (4), 10147-10166. https://doi. org/10.18400/tekderg.492280 [65] Yildiz, Ö. (2021). Correlation Between Spt and Pmt Results For Sandy And Clayey Soils. Eskiehir Technical University Journal of Science and Technology A - Applied Sciences and Engineer­ing, 22 (2) , 175-188. https://doi.org/10.18038/ estubtda.896491 [66] Cabalar, A. F., Karabas, B., Mahmutluoglu, B., & Yildiz, Ö. (2021). An IDW-based GIS application for assesment of geptechnical characterization in Erzincan Turkey. Arab J Geosci 14:2129. https:// doi. org/ 10. 1007/s12517- 021- 08481-6 [67] Shahin, M.A., Jaksa, M.B., & Maier, H.R. (2001). Artifcial Neural Network applications in geotech­nical engineering. Australian Geomechanics Jour­nal 36(1): 49–62 32. Acta Geotechnica Slovenica, 2022/2 M. C. Acar and T. Hakan: SPT-based soil-liquefaction models using nonlinear regression analysis and artificial intelligence techniques SPT-BASED SOIL-LIQUE­FACTION MODELS USING NONLINEAR REGRESSION ANALYSIS AND ARTIFICIAL INTELLIGENCE TECHNIQUES Mehmet Cemal Acar (corresponding author) Vocational College of Kayseri University, Department of Construction Kayseri, Turkey E-mail: acarc@kayseri.edu.tr MODELI UTEKO°INJENJA ZEMLJIN NA OSNOVI SPT Z UPORABO REGRESIJSKE ANALIZE IN TEHNIK UMETNE INTELIGENCE Tay Hakan Civil Engineer (Geotechnic) DS˝ 12. Bge MlKayseri, Turkey https://doi.org/10.18690/actageotechslov.19.2.33-45.2022 liquefaction, standard penetration test (SPT), ANN, uteko°injenje, standardni penetracijski preizkus (SPT), ANFIS, NMRA ANN, ANFIS, NMRA Saturated, cohesionless soils can temporarily lose their shear strength due to increased pore-water pressure under the e˙ect of repetitive dynamic loads such as earthquakes. ˛is event is defned as soil liquefaction and causes signifcant damage to structures. ˛e liquefaction potential of soils depends on many soil parameters obtained in the feld and from laboratory tests. In this study new models have been developed to estimate the liquefaction potential of cohesionless soils. For this purpose, 837 soil data sets were collected to calculate the liquefaction potential with nonlinear multiple regression and artifcial intelligence in the cities of Kayseri and Erzincan. ˛e models based on Nonlinear Multiple Regression Analysis, Artifcial Neural Networks, and Adaptive Neuro-Fuzzy-Inference System techniques were compared with the results of the simplifed method. Determination coeycients (R2) and various error rates were calculated for the performance-evaluation criteria of the models. ˛e proposed ANN model e˙ectively found the complex relationship between the soil and the input parameters and predicts the liquefaction potential more accurately than other methods. It has an overall success rate of 90 percent and the lowest mean absolute error rate of 0.024. With the improvement of existing methods, new models have been introduced to estimate the liquefaction probability of soils. Zasi°ene nekoherentne zemljine lahko zaradi pove°anega pritiska vode v porah pod vplivom ponavljajo°ih se dina­mi°nih obremenitev, kot so potresi, za°asno izgubijo svojo strižno trdnost. Ta primer je opredeljen kot uteko°injenje zemljine in povzro°i znatno škodo na konstrukcijah. Poten­cial uteko°injanja zemljin je odvisen od številnih parame­trov zemljin, pridobljenih s terenskimi in laboratorijskimi preiskavami. V pri°ujo°i študiji so bili razviti novi modeli za oceno potenciala uteko°injenja nekoherentnih zemljin. V mestih Kayseri in Erzincan je bilo zbranih 837 nizov podat­kov o zemljinah za izra°un potenciala uteko°injenja z neli­nearno multiplo regresijo in umetno inteligenco. Modele, ki temeljijo na tehnikah nelinearne multiple regresijske analize (NMRA), umetnih nevronskih mrež (ANN) in sistema prilagodljivega nevro-mehkega sklepanja (ANFIS), smo primerjali z rezultati poenostavljene metode. Za kriterije ocenjevanja uspešnosti modelov so bili izra°unani determinacijski koefcienti (R2) in razli°ne stopnje napak. S predlaganim modelom ANN smo našli kompleksno razmerje med zemljino in vhodnimi parametri ter napo­vedali potencial uteko°injenja natan°neje kot z drugimi metodami. Model ANN ima skupno stopnjo uspešnosti 90 odstotkov in najnižjo srednjo absolutno stopnjo napake 0,024. Z izboljšanjem obstoje°ih metod so bili uvedeni novi modeli za oceno verjetnosti uteko°injenja zemljin. M. C. Acar and T. Hakan: SPT-based soil-liquefaction models using nonlinear regression analysis and artificial intelligence techniques 1 INTRODUCTION When natural disasters are evaluated in terms of loss of life and property, earthquakes come frst. An earthquake has negative e˘ects on building structures. In particular, when the saturated sandy and silty soils are liquefed, they cannot bear the weight of the structures standing on them, which causes the structures to sink and tilt [1]. ~e liquefaction of soils can be expressed as the liquid-like behaviors of saturated, cohesionless or low-cohesive soils that lose their shear strength because of the vibra­tions of cyclic, earthquake shock waves. Liquefaction is the increase of the pressure of the water in the soil void spaces (pores) and the deterioration of the soil’s structure under repeated loads from the e˘ect of the earthquake. ~e increase in pore-water pressure reduces the e˘ective stress in the soil and then leads to a loss of shear strength and the soil starts to act like a liquid. When earthquakes with a moment magnitude (Mw) greater than 5 are examined, liquefaction occurs most frequently on loose, saturated sandy and silty soils. ~ese deformations caused buildings to collapse or be severely damaged. For example, in the Niigata, Japan earthquake with a moment magnitude of 7.5 occurred in 1964, concrete buildings sank and tilted laterally. In 1995, in the 7.2 magnitude earthquake in Kobe, Japan, bridges, buried pipelines, port facilities, the retaining walls on the coasts were damaged by tilting and the buildings sank due to liquefaction. Similarly, a 7.5 magnitude earthquake occurred in Glck-Marmara, Turkey in 1999. Many structures sank into the soil or tilted to one side [1-3]. Assessments of liquefaction risk started with the 1964 Niagata earthquake and became much more important for the 1971 San Fernando, 1985 Mexico City, 1989 Loma Prieta, 1995 Kobe and 1999 Marmara earthquakes. Laboratory and feld tests are used in the liquefaction assessment based on the simplifed method. Competent people should prepare the samples representing the feld conditions to obtain the correct results from laboratory experiments. It is a challenging and laborious job in practice. Damage oen occurs during the collection of the samples from the feld and the transportation and preparation for the experiment. ~erefore, a determina­tion of the liquefaction potential according to just the results of a laboratory experiment causes errors. Instead, the results of the Standard Penetration Test (SPT) carried out in the feld are oen used to estimate a soil's liquefaction resistance [4]. Some factors a˘ecting liquefaction include ground water, earthquake moment size, soil type, corrected soil SPT penetration resistance (N60) value, relative density (Dr), depth of the obtaining SPT N60 value (d), fnes content (FC), which is defned as the ratio of soils passing a No. 200 (75 m) sieve, average grain diameter (D50), unit weight (UW), groundwater level (GWL), e˘ective stress ('vo), total stress (vo), peak ground acceleration (amax) and depth of the earthquake from the surface [1, 2, 5-10]. Seed and Idriss [4] carried out the frst stud­ies on liquefaction based on SPT data. ~ey suggested that liquefaction could be estimated with graphs and equations, depending on SPT. A liquefaction assessment using SPT data, which was developed by Seed and Idriss, is referred to as the “simplifed method”[4]. Versions of this method improved by other authors are used world­wide [11],[12],[13]. ~e Turkish Building Earthquake Code (TBEC-2018) accepts the "simplifed method", which consists of the empirical equations dependent on the SPT published by Seed and Idriss[4] as the standard method for soil-liquefaction analyses. ~e cyclic stress approach is used to evaluate the lique­faction potential. In this approach, the Cyclic Stress Ratio (CSR) represents the earthquake load, i.e., the earth-quake's soil liquefaction e˘ect or demand. Depending on the results of the SPT test in the feld, the resistance of the soil to liquefaction is represented by the Cyclic Resistance Ratio (CRR). ~e fact that the liquefaction factor of safety, i.e., the CRR over CSR ratio, is less than 1.10 usually means that soil will liquefy according to TBEC-2018 [38]. ~e cyclic-stress approach in assessing the liquefaction potential expresses both the earthquake e˘ect (CSR) and the soil-liquefaction resistance (CRR) in cyclical stresses. ~e cycle number for the CSR, which is a function of the duration of the earthquake movement, is proportional to the magnitude of the earthquake. ~e cyclic liquefaction resistance (CRR) is obtained in the laboratory by cyclic triaxial and simple shear tests or most oen in the feld by the SPT. ~e CRR is expressed in terms of the number of cycles required for the occur­rence of collapsing in a soil exposed to cyclic shear stress at a certain level. However, the CRR is a˘ected by the stress and unit deformation history, age, and soil texture, which are disturbed during sampling and are very diycult to simulate in the laboratory. ~erefore, feld tests are preferred for a liquefaction assessment. ~e SPT is a widely used feld test used to calculate the CRR of the soil. In recent years the number of scientifc studies using numerical methods based on statistical and artifcial intelligence techniques for estimating the liquefaction of soils has increased [9, 10, 14-23]. Finn, Dowling and Ventura[22] developed methods that estimate the liquefaction potential and lateral expansion displace­ments. Boulanger and Idriss[13] studied the probability of triggering liquefaction based on SPT. In their study, 34. Acta Geotechnica Slovenica, 2022/2 M. C. Acar and T. Hakan: SPT-based soil-liquefaction models using nonlinear regression analysis and artificial intelligence techniques they obtained the maximum probability approach and the liquefaction trigger correlation related to SPT. Kera­matikerman, Chegenizadeh, and Nikraz [24] conducted a series of repeated triaxial experiments to determine the e˘ect of fy ash on the liquefaction resistance of sands, and they observed that the resistance to liquefaction increases with the increasing ash ratio and time. Yang et al. developed an SPT-based empirical equation to assess sand liquefaction [19]. Rahman and Siddiqua [21] estimated the liquefaction resistance of soils using the standard penetration test, cone penetration test, and shear wave velocity data for the cities of Dhaka, Chittagong and Sylhet in Bangladesh. ~e e˘ects of FC on the liquefaction of soils were also investigated [25]. Anwar et al. obtained a model to fnd the CRR for MRA-based soil-liquefaction analysis using SPSS and the MATLAB program [26]. Fei-hong[27] investigated the statistical relationship between the liquefaction index and the depth using SPT data to assess soil liquefac­tion in the port area of Tianjin city, and they showed a signifcant relationship between the liquefaction index, the depth, and the SPT N-value. Another study claimed that liquefaction is a complex ground-degradation problem involving soil and earthquake parameters, and ground deformations caused by liquefaction should be investigated by nonlinear methods [28]. In a study inves­tigating fuzzy neural network models for the prediction of liquefaction, integrated fuzzy neural network models were developed to evaluate the liquefaction potential [5]. Muduli and Das developed an empirical model using multi-gene genetic programming (MGGP), which is an SPT-based artifcial intelligence technique, to determine the CRR of the soil [23]. In a study estimating the safety coeycient against liquefaction with artifcial neural networks, the liquefaction potential of the soils in the adjacent area of Gler-Denizli province was evalu­ated, and the safety coeycient against liquefaction was estimated with the help of ANN [29]. In this study, standard penetration tests were carried out in 63 drill holes in Erzincan and 60 drill holes in Kayseri. Seed and Idriss’s simplifed liquefaction analysis [4] was used to determine the liquefaction safety coeycient and the e˘ects of various soil parameters on calculating the CRR were investigated. A data set containing these parameters and the liquefac­tion safety coeycients was prepared. First, a quadratic nonlinear multiple regression model (NMRA) that predicts the soil liquefaction and refects the nonlinear behavior of the soil was developed with this data set. Consequently, models that predict the liquefaction of the soils were created with Artifcial Neural Networks (ANN) and then with the Adaptive Neuro-Fuzzy Infer­ence System (ANFIS) using the same training and test data. Randomly selected training and test data were used in the development of the NMRA, ANN, and ANFIS models. To develop the best model, CRRact values obtained as a result of a simplifed liquefaction analysis were compared with the predicted CRRpred values using the NMRA, ANN, and ANFIS models. ~is study examines the estimation methods against the liquefaction hazard that an earthquake can cause. By determining the liquefaction potentials and comparing the estimation methods, the soil improvements and geotechnical designs will be more secure. ~ey can help to prevent the devas­tating consequences of earthquake-induced liquefaction. 2 METHOD OF MODELLING ~e standard penetration test is widely used in the calcu­lation of liquefaction analysis. SPT is a simple and rela­tively low-cost feld test for the evaluation of liquefaction potential due to easy data acquisition, the presence of a database prepared from the data obtained in previous earthquakes, and revealing a good correlation of these data with new earthquakes. Ref [4] proposed equation (1) for the liquefaction analysis. (1) where amax = peak horizontal ground acceleration on the soil surface (m/s2) g = acceleration due to gravity (m/s2) vo = total vertical soil stress (kPa) 'vo = e˘ective vertical soil stress (kPa) rd = stress-reduction coeycient from equations (2) and (3) ~e largest (CSR) in the formula is the ratio of the mean shear stress (av = 0.65 * max) to the e˘ective vertical stress. ~e e˘ective stress-reduction coeycient rd is a value that considers the fexibility of the soil column (e.g., rd = 1 corresponds to rigid mass behavior). (2) (3) Ref [12] proposed equation (4) for calculating the recur­rent resistance ratio (CRRM7.5) for clean sands with an FC of less than 5 % and earthquakes with a magnitude of Mw = 7.5. ~e next step is to fnd the dynamic cyclic resistance ratio (CRR) for the ground, based on the calculated clean sand equivalent. M. C. Acar and T. Hakan: SPT-based soil-liquefaction models using nonlinear regression analysis and artificial intelligence techniques (4) CRRM7.5 = cyclic resistance ratio to soil liquefaction for Mw = 7.5 earthquake Considering the e˘ect of the FC on the liquefaction resistance, the corrected SPT-N values used in the lique­faction analysis are suggested to be corrected as follows. Ref [12] proposed using the (N1)60 value aer converting to the clean-sand equivalent (N1)60cs . ~ey wanted to reduce the impact of FC on the soil on the CRR. Equa­tion (5) is given as follows. (5) Where  and  are the fnes-content correction coef­fcients. ~e (N1)60cs value is calculated using equations (6), (7) and (8) according to the FC ratio. (6) (7) (8) where SPT-Nfeld is the value adjusted to 60 % of the energy ratio and (N1)60cs is the number of SPT blow-count values with the fnes-content correction. Liquefaction occurs when the CRR, which shows the soil's resistance to liquefaction, exceeds the liquefaction resistance (CSR) caused by earthquakes. If this situation is explained in terms of safety factor, the Safety Coef­fcient (FS) is given by equation (9). (9) FS  1.1 is considered as there being a liquefaction potential and FS 1.1 is considered as there being no liquefaction potential [38]. ~e equations and curves given for the calculation of CRR are valid for an earthquake with a moment magnitude M = 7.5. ~e Figure 1. Site layout and borehole location plan in Kayseri province. 36. Acta Geotechnica Slovenica, 2022/2 M. C. Acar and T. Hakan: SPT-based soil-liquefaction models using nonlinear regression analysis and artificial intelligence techniques earthquake-magnitude correction factor specifed in equations (10) and (11) is proposed to use in di˘erent earthquake magnitudes. (10) MDF = earthquakmagnitude correction factor (11) M = earthquake moment magnitude 2.1 Study Area Description ~e analysis was performed using data from two di˘erent cities (Kayseri and Erzincan). In Kayseri, there are generally sand and silt soil layers in the area under investigation (approximately 1.5 million square meters) (Figure 1). However, the soil properties of the region have variable soil conditions, and it is silty sand in some places and silty clay with sand inter bands at some other sites. ~e 60 borehole drillings, shown in Figure 1, were performed in the study area, and SPT was achieved every 1.5 m in drillings between SK1-SK48 and SK49-SK60. ~e peak horizontal ground acceleration at the ground surface (amax) value can be expressed in gravitational acceleration (g). In the study area of Kayseri, the amax value changes between 0.190 g and 0.200 g (TBEC-2018). ~e earthquake that is thought to a˘ect Kayseri Province and its surroundings is the movement that might occur in the Ecemi fault, which is a strike slip fault. ~e second study area is in the Erzincan plain in the city center of Erzincan. ~e soil types in Erzincan are generally non-plastic, silty sand and clayey sand. ~e peak ground acceleration, amax value changes between 0.600 g and 0.615 g in the Figure 2. Bore holes in the study area in Erzincan province[33]. Note: SK = Borehole M. C. Acar and T. Hakan: SPT-based soil-liquefaction models using nonlinear regression analysis and artificial intelligence techniques study area in Erzincan. ~is region is located on the KAFZ (North Anatolian Fault Zone), which is the most e˘ective fault zone of Turkey. ~e earthquake parameters were calculated separately using the geographical location data entry for Erzincan and Kayseri with the interactive web application (https://deprem.afad.gov.tr). According to the soil class in the study areas, the earthquake ground-motion level with a probability of exceeding 10 % in 50 years (recurrence period of 475 years) was taken into account (this is called DD-2, TBEC-2018). According to the earthquake ground-motion level, the peak acceleration values varied between 0.600 g and 0.615 g and 0.190 g and 0.200 g in the study area of Erzincan and Kayseri, respectively (TBEC-2018). According to these acceleration values, it was estimated as Mw 7.5 for Erzincan and Mw 6 for Kayseri. In the estima­tion of these moment-magnitude values, the approaches given in the literature were used [39]. In the city of Erzincan, 63 borehole drillings with depths of 1.5 m to 20 m, shown in Figure 2, were made. ~e soil parameters and SPT data were collected, and soil profles were created for these drillings. 3 PROCESSING AND ANALYSIS OF DATA ~is study explains the CRR values obtained from Simplifed Liquefaction Analysis [4], NMRA, ANFIS, and ANN methods for Kayseri and Erzincan. Groundwater levels varied between 1.7 and 2.8 m in the study area in Kayseri. ~e Unit Weight Test to determine the mass properties of the soil, the Water Content Test to determine the amount of water in the unit volume of the ground, the Liquid Limit and Plastic Limit Tests to deter­mine the consistency characteristics, the Sieve Analysis Test to determine the grain diameter and distribution of the soil and Hydrometer test to determine the FC were carried out in the samples taken from the investigated area. As the result of the tests, it was observed that the unit-weight values varied between 15 and 20 kN/m3, the FC ranged from 12 to 45 %, and the water contents ranged between 13 % and 47 %. ~e soils’ liquid limit (LL) values varied between 30 % and 47 %, and the plas­tic limit (PL) values ranged from 20 % to 27 %. In Erzincan, 63 boreholes were drilled in 16 di˘erent locations with depths ranging from 1.5 to 20 m. ~e soil types are SM (silty sands), SC (clayey sands), and CL (inorganic clay of low plasticity), which are character­ized as primarily coarse-grained soils. ~erefore, tests were made on the samples and the following results were obtained; the unit-weight values varied between 16.93 and 19.96 kN/m3, FC ranged from 12 to 77 %, and the water-content (w) value varied between 10 and 30 %. ~e soil types in the study area were generally non-plastic (NP) according to the PL and LL test results. According to the results of the SPT obtained from both regions and the laboratory experiments, 837 data sets were obtained. ~e reason for choosing these data is that the groundwater table is close to the ground surface and the soil properties also show the liquefaction potential in both regions. ~ese data were randomly divided into two groups: a training group composed of 70 % (586) and a test group consisting of 30 % (251) of the data. Data of the parameters used in CRR calculation were the earthquake magnitude (Mw), depth (d), corrected soil SPT penetration resistance (N60), saturated Unit Weight (UW), peak ground acceleration (amax), fnes content (FC), cyclic stress ratio (CSR), groundwater level (GWL), total stress (vo) and average grain diameter (D50). ~ere are the e˘ects of 9 di˘erent independent variables on the calculation of CRR. In liquefaction, the groundwater level is generally crucial up to the frst 3 m from the surface. Although liquefaction rarely occurs in environments where the groundwater level is deeper than 10 m from the surface, liquefaction is not expected in environments where the groundwater level is deeper than 20 m, in general [36]. In a complex hydrogeological environment, the groundwater level is variable due to hydraulic transitions that a˘ect the hydraulic properties of the soil. Moss et al. (2017) investigated the e˘ect of the groundwater level on the liquefaction potential and the e˘ect of changing the depth of the water table on the liquefaction according to seasonal variations. ~e water level rises to its maximum during the rainy season due to rain. ~e study highlights the need for seasonal liquefaction-sensitivity studies [37]. ~e groundwater levels varied between 1 and 2.7 m and 1.7 and 2.8 m in the study area of Erzincan and Kayseri, respectively. ~e groundwater table is assumed to be at the surface for the worst-possible scenario for both regions in determining the liquefaction hazard by considering seasonal and global climate change. Besides, since the groundwater-depth values measured in the feld show minor local variations according to the regions, these parameters did not make a signifcant di˘erence in the training of the prediction models compared to the e˘ect of other parameters. For this reason, it was not preferable to use the groundwater-level depth parameters as variables in the models. In estimating the CRR, N60, d, FC, D50, UW, GWL, e˘ec­tive stress ('vo), and total stress (vo) parameters were considered. ~e soils are completely saturated below the water table since the groundwater layer is assumed to be at the surface. Saturated unit weights were used 38. Acta Geotechnica Slovenica, 2022/2 M. C. Acar and T. Hakan: SPT-based soil-liquefaction models using nonlinear regression analysis and artificial intelligence techniques for a point below the groundwater table to calculate the e˘ective unit weight. However, since the soil depth and UW are used to calculate the total stress values, only the depth and saturated unit weight represent the total stress to avoid over-learning in the estimation methods. In addition, a Variance Infuence Factor (VIF) analysis was performed to see the e˘ect of independent variables on CRR. All the estimation models used in this study are based on the fve most infuential variables for the saturated condition of the soils as a result of a VIF analysis. ~ese are, namely, the SPT value (N60), fnes content (FC), saturated unit weight (UW), depth from which SPT is obtained (d), and average grain diameter (D50) parameters. Regression analysis coeycients, T-test values, and VIF values, which were obtained with the analysis performed to determine the relationship between the dependent and independent variables, are given in Table 1. In Table 1, the N60, UW, D50, d, and FC parameters are preferred as independent variables, since the VIF values were less than 5. It is clear that the t values obtained were not within the range -1.645 < t < 1.645, which were determined for tcritical = 1.645. ~us, the N60 , UW, D50 , d, and FC parameters were signifcant in the CRR estimation and were used in the analysis. Table 1. Results of regression analysis according to t test and VIF analysis. Independent Regression t VIF variablescoeycients Constant 2.7806 6.88 N60 0.0212426 26.39 1.581 FC 0.17645 2.03 1.268 UW 0.16398 7.08 1.795 d 0.0018631 2.24 1.112 D50 0.6052 2.78 1.257 tcritical = 1.645 Table 2. Data statistics. Parameters n Min Max Mean Standard deviation N60 837 2 53 16.86 9.873 FC 837 0.12 0.77 0.36 0.076015 UW 837 16.93 19.96 18.16 0.36116 d 837 1.50 40.50 11.23 7.86592 D50 837 0.01 0.334 0.042 0.023233 CRR 837 0.084 1.24 0.30 0.1426768 It was reported that the test performance of fuzzy logic-based models such as ANFIS decreases with an increase in the number of independent variables [5]. ~erefore, the number of independent variables was limited to 5 in this study. Table 2 shows the statistical data of this dependent (CRR) and the independent variables. 3.1 Performance Criteria In estimating the CRR value, MAE, MSE, RMSE, MARE, and R2 are taken into account to compare the performance of the models. ~e model error rate occurs because it does not fully represent a proper relationship between the predicted and the actual parameters. As a result of this incomplete relationship, di˘erent error-rate indices can be expressed. ~e mean absolute error (MAE) is the measured di˘erence between two vari­ables. ~e MAE is also the average horizontal distance between each data point and the best-ft line. Since the MAE value is easily interpretable, it is frequently used in regression and artifcial intelligence techniques. ~e MAE value can vary from zero to infnity. ~e mean square error (MSE) measures the performance of the model, the estimator, telling how close the prediction curve is to a set of points. When the MSE value is zero, the model has the best-possible performance. ~e RMSE (root-mean-square error) is the standard deviation of the estimation errors. ~e RMSE is a measure of the distribution of residues. ~e RMSE value can range from zero to infnity. A zero RMSE value means that the model made no errors. ~e MARE expresses the di˘erence between the estimated value and the observed value. ~e MARE is a non-negative error rate that can take a value from zero to infnity. When the MARE value is zero, the considered model has the best-possible performance. ~e performance criteria used for model evaluation in this study are given in Table 3. ~ese are (R2), MAE, MARE, MSE and RMSE. Here, the value of R2 indicates the closeness of our model (as a percentage) to the real values. In Table 3, CRRact , CRRact , CRRpred , CRRpred are the real values of CRRact , calculated by simplifed liquefaction analysis, the calculated real mean CRRact , the predicted CRRpred , and the predicted mean CRRpred , respectively. Equation (12) is used to normalize the data to transfer to the MATLAB program. (12) Here, Xn is normalized data, X0 is original data, Xmin is minimum data and Xmax is maximum data. All data were scaled between 0 and 1. M. C. Acar and T. Hakan: SPT-based soil-liquefaction models using nonlinear regression analysis and artificial intelligence techniques Table 3. Performance-evaluation criteria. Evaluation Defnition criteria Coeycient of determi­nation (R2) Mean ab­solute error (MAE) Mean abso­lute rela­tive error (MARE) Mean square er­ror (MSE) Root mean square er­ror (RMSE) 3.2 Nonlinear Multiple Regression Analysis (NMRA) model Nonlinear multiple regression analysis (NMRA) is used to detect two or more correlations. NMRA is a statistical method that can reveal the relationship between more than one independent variable and a single dependent variable, make predictions, and create a mathematical model. In this study, quadratic regression was used to estimate the CRR value. ~e basic equation of the regres­sion model is relatively simple, as given by Equation 13. CRR represents the dependent variable; N60 , FC, UW, d, and D50 are the independent variables. ~e ability of the estimations to give reliable results depends on the coef­fcient of determination (R2) being the largest value and the error rates being the smallest value. ~en, with the help of the SPSS program, various func­tions were tested with these independent variables, and the best ft for the distribution is the nonlinear quadratic equation. ~e quadratic NMRA equation was chosen, which gave the highest R2 and the lowest RMSE values. ~e nonlinear regression equation obtained from the NMRA analysis is shown in Equation 13. ~us, the R2 was 0.718 for the training data and 0.681 for the test data. Other error statistics of the NMRA model are shown in Table 6. As shown in Figure 3, the CRRact and CRRpred values were close to each other. However, as shown in Figure 3, the CRR values diverged from the calculated CRR values aer 0.80, which results in a reduction of the determination coeycient. (13) Figure 3. Comparison of predicted and actual CRR values for NMRA model in training and test data (Nonlinear Multiple Regression Analysis). 40. Acta Geotechnica Slovenica, 2022/2 M. C. Acar and T. Hakan: SPT-based soil-liquefaction models using nonlinear regression analysis and artificial intelligence techniques 3.3 Artifcial Neural Network (ANN) model ~e estimation method, called ANN (Artifcial Neural Networks), is the most well-known and widely used method among the artifcial intelligence techniques. ~is method estimates the dependent variable by fnding linear or nonlinear relationships between the parameters that represent many independent variables. In this technique, the working system imitates the human brain. ANN makes routing with multi-layer sensor networks and can learn and generalize between the input and output layers. ~e ANN structure consists of an input layer, an output layer, many hidden layers, and a large number of neurons corresponding to each independent variable. An ANN is very successful in fnding nonlinear relationships between independent variables about the dependent variable. ~e output layer also corresponds to the predicted dependent variable. ~e system updates the weight values, moving from the output layer to the input layer, and the error value is minimized [5,10]. ~e CRR was estimated with the ANN model. In the predic­tion model, the input parameters are N60 , FC, UW, d, D50 , and the output parameter is the CRR (Figure 4). ~e feed-forward ANN model inputs with fve variables consisting of N60 , FC, UW, d and D50 , and a single output system CRR was obtained, as shown in the diagram in Figure 4. In the training of the models, a random selection of 586 parameters was used, and 251 parameters were used to test the prediction model's performance. First, the most appropriate and widely used tansig, logsig and purelin functions from 11 member functions were used in the multi-layered ANN method in MATLAB [34]. ~e input data were trained with the Levenberg-Marquardt algorithm, due to its ease of use, convergence rate and predictive success in linear and nonlinear models. ~e numbers of neurons in a hidden layer ranged from 2 to 10, and the numbers of iterations ranged from 1 to 100. Using a trial-and-error method, a model was determined from the network structures obtained. In this model, the input membership function was logsig, which gives the lowest all error statistic values and the highest Determination Coeycient (R2) value, and its output membership function was purelin. Table 4. Best ANN model for predicting CRR. Membership function Input Output Membership function number Iteration number Logsig Purelin 10 95 Figure 5. Comparison of predicted CRR and actual CRR values for ANN model. M. C. Acar and T. Hakan: SPT-based soil-liquefaction models using nonlinear regression analysis and artificial intelligence techniques ANN characteristics of the best ANN model obtained within this layer to obtain the actual output value of the aer various trial-and-error attempts are given in Table ANFIS system [5] 4. All the statistical error values of the best ANN model are given in Table 6. According to Table 6, the error statistic values were within acceptable values. As can be seen in Table 6, and Figure 5, the ANN model revealed successful results. However, it can be seen in Figure 5 that the predictive CRR values diverged between 1.0 and 1.5 from the calculated CRR values. 3.4 Adaptive neuro-fuzzy inference system (ANFIS) model ~e adaptive ANN-based fuzzy inference system (ANFIS) is one of the essential artifcial intelligence techniques that can optimize parameters with an infer­ence system. ANFIS provides for the optimization of rule-base and membership function values to model systems with known input and output values with fuzzy logic. ~e optimization process involves th learning methods of ANN. In this way, fuzzy systems, which normally cannot learn, gain a learning ability for the data sets to be modeled. ANFIS uses the backpropaga­tion method, as a learning method, or a combination of the backpropagation method and the least-squares estimation method. ~e ANFIS architecture consists of six layers. ~e frst layer (input layer) transmits the incoming input signals to the other layers. ~e second layer is called the fuzzy layer, the third layer is the rule layer, and the fourth layer is the normalization layer. ~e fh layer is the annotation layer, and fnally, in the sixth layer, the values from the annotation layer are aggregated In the ANFIS model, since the increase in the member­ship function numbers, as mentioned above, decreased the performance of the ANFIS model, the input membership function numbers 2 and 3 were taken. Aer determining the Gaussian membership function (gaussmf) and the triangular (trimf) membership function as the input membership functions and the constant and linear functions as the output functions, the best ANFIS model was determined by trial and error in iteration numbers ranging from 1 to 5. Depend­ing on the type of input and output functions, Gaussmf-constant, Gaussmf-linear, trimf-constant and trimf-linear combinations were determined. ~e lowest errors criteria and the highest R2 are used to select the best model among the four di˘erent ANFIS combinations. ~e ANFIS features of the best ANFIS model obtained from various trial-and-error attempts are given in Table 5. ~e input membership function is Trimf, and the output membership function is linear. Allthe error-statistics values of the best ANN model and successful results of the ANFIS model are given in Table 5, Figure 6 and Table 6. Table 5. Best ANFIS model for CRR prediction. Membership function Input Output Membership function number Iteration number Trimf Linear 2 4 Figure 6. Comparison of predicted CRR and actual CRR values for the ANFIS model. 42. Acta Geotechnica Slovenica, 2022/2 M. C. Acar and T. Hakan: SPT-based soil-liquefaction models using nonlinear regression analysis and artificial intelligence techniques However, as can be seen in the scatter chart (Figure 6), the CRRpred values between 1.0 and 1.5 diverged from the CRRact values. ~is situation, which decreases the R2 coeycient value, indicated that the model failed between 1.0 and 1.5 values. 4 RESULTS AND DISCUSSION As shown in Table 6, the accuracy of the model's results was accepted as satisfactory, with a determination coef­fcient greater than 70 % obtained for all the methods. However, considering all the error statistics, the ANN model seems the best among these three methods in terms of R2 value and the lowest MAE, highest MARE, and MSE ratios. (Table 6). ~e results of the models based on the Nonlinear Multiple Regression Analysis (NMRA), Artifcial Neural Networks (ANN), and Adaptive Neural Fuzzy Inference System (ANFIS) were compared with simplifed analysis results in order to develop the best model for estimating the liquefaction of soils. In all models developed using statistical and artifcial intelligence techniques, N60 , FC, UW, d, D50 were used as the input parameters, and the CRR value was estimated as the output parameter. Table 6. Performance statistics of all models. ANN ANFIS NMRA MAE 0.024 0.0350 0.095 MARE 95.214 11.0330 41.900 MSE 0.0019 0.0068 0.018 RMSE 0.0432 0.0822 0.134 R2 0.968 0.885 0.718 MAE 0.034 0.036 0.098 MARE 12.675 10.798 0.441 MSE 0.006 0.010 0.021 RMSE 0.0777 0.1002 0.145 R2 0.901 0.838 0.681 Test Training ~e R2 and various error ratios were calculated by comparing the performance of the models created with the training data. It was concluded that the most suitable model is the ANN model based on the success rates and consistency of the models. Liquefaction analysis was carried out for di˘erent drilling depths and soil proper­ties. ~e liquefaction status was calculated by loading data into the Excel spreadsheet. ~e VIF analysis was performed to see which parameters infuence the CRR calculation. ~e numerical studies in the CRR calcula­tion showed that the N60, FC, UW, d, and D50 parameters a˘ect the CRR estimation aer the VIF analysis. ~us, new models with higher accuracy were produced using these most infuential parameters. ~e determination coeycients were 0.681, 0.838, and 0.901 in the NMRA, the ANFIS, and the ANN methods, respectively. ~ese values show that a desired level of estimation is achieved. When the error criteria in the NMRA, ANN, and ANFIS methods were evaluated, it was observed that the ANN method was superior to the other methods, with an overall success rate of 90 % and the lowest mean abso­lute error rate of 0.024. 5 CONCLUSIONS In this study, a nonlinear multiple regression analysis was performed between the CRR value and the other soil parameters. ~en, the CRR estimation was made with fuzzy logic and artifcial neural networks for the N60 , FC, UW, d and D50 variables, which gave high correlation coeycients. ~e most sensitive parameters to the soil’s liquefaction are d, FC, and N60, while the least sensitive ones are the UW and D50 soil parameters. ~e ANN model used in the CRR estimation has more successful performance criteria when the comparing R2 and errors. ~e ANN model has lower error values and a higher correlation than the NMRA and ANFIS models. Improv­ing the existing methods for predicting the liquefaction of soils and estimating the probability of liquefaction with the new models to be developed will enable civil engineers to take precautions against liquefaction. In addition, this study has demonstrated the successful and rapid use of artifcial intelligence techniques in solving geotechnical problems, especially in modeling nonlinear complex soil behavior, such as liquefaction. ~is study is the basis for further studies on liquefaction. In future studies that can be performed to obtain better results in the calculation of the CRR, the number of model data can be increased by adding new data. Also, new models and higher accuracy results can be obtained using di˘er­ent artifcial intelligence techniques. REFERENCES [1] Yoshida, N., Tokimatsu, K., Yasuda, S., Kokusho, T., and Okimura, T. (2001). Geotechnical aspects of damage in Adapazari city during 1999 Kocaeli, Turkey earthquake. Soils and foundations, 41(4), 25-45. M. C. Acar and T. Hakan: SPT-based soil-liquefaction models using nonlinear regression analysis and artificial intelligence techniques [2] Youd, T. L. (1993). Liquefaction-induced damage to bridges. Transportation Research Record, 1411, 35-41. [3] Finn, W. L., Byrne, P. M., Evans, S., and Law, T. (1996). Some geotechnical aspects of the Hyogo-ken Nanbu (Kobe) earthquake of January 17, 1995. Canadian Journal of Civil Engineering, 23(3), 778-796. [4] Seed, H. B. and Idriss, I. M. Simplifed procedure for evaluating soil liquefaction potential. Journal of Soil Mechanics Foundations Div, 97, No SM9, PROC PAPER 8371, ,(1971) 1249-1273. [5] Rahman, M. S. and Wang, J. Fuzzy neural network models for liquefaction prediction. Soil Dynamics and Earthquake Engineering, 22, 8,(2002) 685-694. [6] Boulanger, R. W. and Idriss, I. M. Probabilistic Standard Penetration Test-Based Liquefaction-Triggering Procedure. Journal of Geotechnical and Geoenvironmental Engineering, 138, 10,(2012) 1185-1195. [7] Anwar, A., Jamal, Y., Ahmad, S. and Z Khan, M. Assessment of liquefaction potential of soil using multi-linear regression modeling. International Journal of Civil Engineering & Technology (IJCIET), 7,(2016) 373-415. [8] Muduli, P. and Das, S. "Evaluation of liquefaction potential of soil based on standard penetration test using multi-gene genetic programming model". In Proceedings of the Indian Geotechnical Confer­ence IGC2016 (Chennai, India,(2016). [9] Kurnaz, T. F. and Kaya, Y. SPT-based liquefaction assessment with a novel ensemble model based on GMDH-type neural network. Arabian Journal of Geosciences, 12, 15,(2019) 456. [10] Sabbar, A. S., Chegenizadeh, A. and Nikraz, H. Prediction of Liquefaction Susceptibility of Clean Sandy Soils Using Artifcial Intelligence Techniques. Indian Geotechnical Journal, 49, 1,(2019) 58-69. [11] Seed, H. and Idriss, I. Ground motions and soil liquefaction during earthquakes: engineering monographs on earthquake criteria, structural design, and strong motion records. MNO-5. Earth­quake Engineering Research Institute, Oakland, CA, (1982). [12] Youd, T. L. and Idriss, I. M. Liquefaction resistance of soils: summary report from the 1996 NCEER and 1998 NCEER/NSF workshops on evaluation of liquefaction resistance of soils. Journal of geotech­nical geoenvironmental engineering, 127, 4,( 2001) 297-313. [13] Boulanger, R. W. and Idriss, I. Probabilistic stand­ard penetration test–based liquefaction–triggering procedure. Journal of Geotechnical Geoenviron-mental Engineering, 138, 10, (2012) 1185-1195. [14] Geyin, M., Baird, A. J. and Maurer, B. W. Field assessment of liquefaction prediction models based on geotechnical versus geospatial data, with lessons for each. Earthquake Spectra, (2020) 1-26. [15] Farrokhi, F., Firoozfar, A. and Maghsoudi, M. S.Evaluation of liquefaction-induced lateral displacement using a GMDH-type neural network optimized by genetic algorithm. Arabian Journal of Geosciences, 13, 1,(2020) 4. [16] Moayedi, H., Mosallanezhad, M., Rashid, A. S. A., Jusoh, W. A. W. and Muazu, M. A. A systematic review and meta-analysis of artifcial neural network application in geotechnical engineering: theory and applications. Neural Computing Appli­cations, (2018) 1-24. [17] Kirba, . Short-term multi-step wind speed predic­tion using statistical methods and artifcial neural networks. Sakarya University Journal of Science, 22, (2018) 24-38. [18] Asvar, F., Shirmohammadi Faradonbeh, A. and Barkhordari, K. Predicting potential of controlled blasting-induced liquefaction using neural networks and neuro-fuzzy system. Scientia Iranica 25, 2,(2018) 617-631. [19] Yang, Y., Chen, L., Sun, R., Chen, Y. and Wang, W. A depth-consistent SPT-based empirical equation for evaluating sand liquefaction. Engineering Geology, 221, (2017) 41-49. [20] enol, Ü. and Musayev, Z. Estimating wind energy potential by artifcial neural networks method. Bilge International Journal of Science and Technol­ogy Research, 1, 1, (2017) 23-31. [21] Rahman, M. Z. and Siddiqua, S. Evaluation of liquefaction-resistance of soils using standard penetration test, cone penetration test, and shear-wave velocity data for Dhaka, Chittagong, and Sylhet cities in Bangladesh. Environmental Earth Sciences, 76, 5, (2017) 207. [22] Finn, W. D. L., Dowling, J. and Ventura, C. E. Evalu­ating liquefaction potential and lateral spreading in a probabilistic ground motion environment. Soil Dynamics and Earthquake Engineering, 91, (2016) 202-208. [23] Muduli, P. K. and Das, S. K. Model uncertainty of SPT-based method for evaluation of seismic soil liquefaction potential using multi-gene genetic programming. Soils and Foundations, 55, 2, (2015) 258-275. [24] Keramatikerman, M., Chegenizadeh, A. and Nikraz, H.Experimental study on e˘ect of fy ash on liquefaction resistance of sand. Soil Dynamics and Earthquake Engineering, 93,(2017) 1-6. [25] Maurer, B. W., Green, R. A., Cubrinovski, M. and Bradley, B. A. Fines-content e˘ects on liquefaction 44. Acta Geotechnica Slovenica, 2022/2 M. C. Acar and T. Hakan: SPT-based soil-liquefaction models using nonlinear regression analysis and artificial intelligence techniques hazard evaluation for infrastructure in Christch­urch, New Zealand. Soil Dynamics and Earthquake Engineering, 76, (2015) 58-68. [26] Anwar, A., Jamal, Y., Ahmad, S., Z Khan, M. and Publication, I. Assessment of liquefaction potential of soil using multi-linear regression modeling. International Journal of Civil Engineering & Tech­nology (IJCIET), 7,(2016) 373-415. [27] Fei-hong, G. Evaluation of Soil Liquefaction in Harbor District in Tianjin City. ~e Open Civil Engineering Journal, 10, (2016) 293-300. [28] Siyahi, B., Akbas, B. and Dogan Onder, N. "Evalua­tion Of Liquefaction-Induced Lateral Spreading By A Neural Network (Nn) Model". In Proceedings of the 14th World Conference on Earthquake Engi­neering (Beijing, China, 2008). [29] en, G., Akyol, E. and Firat, M. Sivilamaya Kari Gvenlik Katsayisinin Yapay Sinir A lari le Tahmin Edilmesi: Denizli-Gler Örne i. Selçuk Üniversitesi Mhendislik, Bilim Ve Teknoloji Dergisi, 22, (2007) 177-184. [30] Kramer, S. L. Geotechnical Earthquake Engineer­ing. (United Stated of America. Prentice Hall) 1996. [31] Snmezer, Y. B., Çeliker, M. and Kilinç, M. Y. Kirikkale li Bahçelievler ve Fabrikalar Mahal­lelerinin Sivilama Potansiyelinin Co raf Bilgi Sistemlerinde Analizi. International Journal of Engineering Research and Development, 4, 1, (2012) 33-40. [32] Seed, H. B. and Idriss, I. M. Ground motions and soil liquefaction during earthquakes. (Berkeley. Earthquake Engineering Research Institute) 1982. [33] Duman, E. S. Erzincan il merkezi ve çevresindeki zeminlerin standart penetrasyon deneyi verileri kullanilarak sivilama potansiyelinin belirlenmesi. (Determination of the liquefaction potential of soils in and around the centre of the city of Erzin-can using standard penetration-test data) Karad­eniz Teknik Üniversitesi/Fen Bilimleri Enstits, 2013. [34] Beale, M. H., Hagan, M. T. and Demuth, H. B. J. T. M.Neural network toolbox™ user's guide,(2010). [35] Subai Duman, E. Erzincan il merkezi ve çevres­indeki zeminlerin standart penetrasyon deneyi verileri kullanilarak sivilama potansiyelinin belirlenmesi (Determination of the liquefaction potential of soils in and around the centre of the city of Erzincan using standard penetration-test data). MSc. ~esis, Karadeniz Technical University Science Institute, Trabzon, Turkey, 2013. [36] Youd, T. L. 1984. Geological e˘ects-liquefaction and associated ground failure. Geological and Hydrogeological Hazards Training Program, United States Geological Survey Open-File Report 87-76, 210-232. [37] Moss RES, Baise LG, Zhu J, Kadkha D (2017) Examining the discrepancy between forecast and observed liquefaction from the 2015 Gorkha, Nepal, earthquakes. Earthq Spectra 33(S1): S73– S83. https://doi.org/10. 1193/120316EQS220M [38] General Directorate for Foundations, Turkish Building Earthquake Code, TBEC-2018, Ankara, Turkey, (2018). [39] Wang, B. (2020). Geotechnical investigations of an earthquake that triggered disastrous landslides in eastern Canada about 1020 Cal BP. Geoenviron-mental Disasters, 7(1), 1-13. NAVODILA AVTORJEM NAVODILA AVTORJEM Vsebina ~lanka ^lanek naj bo napisan v naslednji obliki:  Naslov, ki primerno opisuje vsebino °lanka in ne presega 80 znakov.  Izvle°ek, ki naj bo skrajšana oblika °lanka in naj ne presega 250 besed. Izvle°ek mora vsebovati osnove, jedro in cilje raziskave, uporabljeno metodologijo dela, povzetek izidov in osnovne sklepe.  Najve° 6 klju°nih besed, ki bi morale biti napisane takoj po izvle°ku.  Uvod, v katerem naj bo pregled novejšega stanja in zadostne informacije za razumevanje ter pregled izidov dela, predstavljenih v °lanku.  Teorija.  Eksperimentalni del, ki naj vsebuje podatke o postavitvi preiskusa in metode, uporabljene pri pridobitvi izidov.  Izidi, ki naj bodo jasno prikazani, po potrebi v obliki slik in preglednic.  Razprava, v kateri naj bodo prikazane povezave in posplošitve, uporabljene za pridobitev izidov. Prika­zana naj bo tudi pomembnost izidov in primerjava s poprej objavljenimi deli.  Sklepi, v katerih naj bo prikazan en ali ve° sklepov, ki izhajajo iz izidov in razprave.  Vse navedbe v besedilu morajo biti na koncu zbrane v seznamu literature, in obratno. Dodatne zahteve  Vrstice morajo biti zaporedno oštevil°ene.  Predložen °lanek ne sme imeti ve° kot 18 strani (brez tabel, legend in literature); velikost °rk 12, dvojni razmik med vrsticami. V °lanek je lahko vklju°enih najve° 10 slik. Isti rezultati so lahko prikazani v tabe­lah ali na slikah, ne pa na oba na°ina.  Potrebno je priložiti imena, naslove in elektronske naslove štirih potencialnih recenzentov °lanka. Urednik ima izklju°no pravico do odlo°itve, ali bo te predloge upošteval. Enote in okrajšave V besedilu, preglednicah in slikah uporabljajte le standardne ozna°be in okrajšave SI. Simbole fzikalnih veli°in v besedilu pišite poševno (npr. ., T itn.). Simbole enot, ki so sestavljene iz °rk, pa pokon°no (npr. Pa, m itn.). Vse okrajšave naj bodo, ko se prvi° pojavijo, izpisane v celoti. Slike Slike morajo biti zaporedno oštevil°ene in ozna°ene, v besedilu in podnaslovu, kot sl. 1, sl. 2 itn. Posnete naj bodo v katerem koli od razširjenih formatov, npr. BMP, JPG, GIF. Za pripravo diagramov in risb priporo°amo CDR format (CorelDraw), saj so slike v njem vektorske in jih lahko pri kon°ni obdelavi preprosto pove°ujemo ali pomanjšujemo. Pri ozna°evanju osi v diagramih, kadar je le mogo°e, uporabite ozna°be veli°in (npr. v, T itn.). V diagramih z ve° krivuljami mora biti vsaka krivulja ozna°ena. Pomen oznake mora biti razložen v podnapisu slike. Za vse slike po fotografskih posnetkih je treba priložiti izvirne fotografje ali kakovostno narejen posnetek. Preglednice Preglednice morajo biti zaporedno oštevil°ene in ozna°ene, v besedilu in podnaslovu, kot preglednica 1, preglednica 2 itn. V preglednicah ne uporabljajte izpisanih imen veli°in, ampak samo ustrezne simbole. K fzikalnim koli°inam, npr. t (pisano poševno), pripišite enote (pisano pokon°no) v novo vrsto brez oklepajev. Vse opombe naj bodo ozna°ene z uporabo dvignjene številke1. Seznam literature Navedba v besedilu Vsaka navedba, na katero se sklicujete v besedilu, mora biti v seznamu literature (in obratno). Neobjavljeni rezultati in osebne komunikacije se ne priporo°ajo v seznamu literature, navedejo pa se lahko v besedilu, °e je nujno potrebno. Oblika navajanja literature V besedilu: Navedite reference zaporedno po številkah v oglatih oklepajih v skladu z besedilom. Dejanski avtorji so lahko navedeni, vendar mora obvezno biti podana referen°na številka. Primer: »..... kot je razvidno [1,2]. Brandl and Blovsky [4], sta pridobila druga°en rezultat…« V seznamu: Literaturni viri so oštevil°eni po vrstnem redu, kakor se pojavijo v °lanku. Ozna°imo jih s številkami v oglatih oklepajih. Sklicevanje na objave v revijah: [1] Jeluši°, P., Žlender, B. 2013. Soil-nail wall stability analysis using ANFIS. Acta Geotechnica Slovenica 10(1), 61-73. 46. Acta Geotechnica Slovenica, 2022/2 Sklicevanje na knjigo: [2] Šuklje, L. 1969. Rheological aspects of soil mechan­ ics. Wiley-Interscience, London Sklicevanje na poglavje v monografji: [3] Mitchel, J.K. 1992. Characteristics and mechanisms of clay creep and creep rupture, in N. Guven, R.M. Pollastro (eds.), Clay-Water Interface and Its Rheo-logical Implications, CMS Workshop Lectures, Vol. 4, ~e clay minerals Society, USA, pp. 212-244.. Sklicevanje na objave v zbornikih konferenc: [4] Brandl, H., Blovsky, S. 2005. Slope stabilization with socket walls using the observational method. Proc. Int. conf. on Soil Mechanics and Geotechnical Engi­neering, Bratislava, pp. 2485-2488. Sklicevanje na spletne objave: [5] Kot najmanj, je potrebno podati celoten URL. ^e so poznani drugi podatki (DOI, imena avtorjev, datumi, sklicevanje na izvorno literaturo), se naj prav tako dodajo. INSTRUCTIONS FOR AUTHORS Format of the paper ~e paper should have the following structure:  A Title, which adequately describes the content of the paper and should not exceed 80 characters;  An Abstract, which should be viewed as a mini version of the paper and should not exceed 250 words. ~e Abstract should state the principal objectives and the scope of the investigation and the methodology employed; it should also summarise the results and state the principal conclusions;  Immediately aer the abstract, provide a maximum of 6 keywords;  An Introduction, which should provide a review of recent literature and suycient background informa­tion to allow the results of the paper to be under­stood and evaluated;  A ~eoretical section;  An Experimental section, which should provide details of the experimental set-up and the methods used to obtain the results;  A Results section, which should clearly and concisely present the data, using fgures and tables where appropriate;  A Discussion section, which should describe the relationships shown and the generalisations made possible by the results and discuss the signifcance INSTRUCTIONS FOR AUTHORS Podatki o avtorjih ^lanku priložite tudi podatke o avtorjih: imena, nazive, popolne poštne naslove, številke telefona in faksa, naslove elektronske pošte. Navedite kontaktno osebo. Sprejem ~lankov in avtorske pravIce Uredništvo si pridržuje pravico do odlo°anja o sprejemu °lanka za objavo, strokovno oceno mednarodnih recenzentov in morebitnem predlogu za krajšanje ali izpopolnitev ter terminološke in jezikovne korekture. Z objavo preidejo avtorske pravice na revijo ACTA GEOTECHNICA SLOVENICA. Pri morebitnih kasnejših objavah mora biti AGS navedena kot vir. Vsa nadaljnja pojasnila daje: Uredništvo ACTA GEOTECHNICA SLOVENICA Univerza v Mariboru, Fakulteta za gradbeništvo, prometno inženirstvo in arhitekturo Smetanova ulica 17, 2000 Maribor, Slovenija E-pošta: ags@um.si of the results, making comparisons with previously published work;  Conclusions, which should present one or more conclusions that have been drawn from the results and subsequent discussion;  A list of References, which comprises all the refer­ences cited in the text, and vice versa. Additional Requirements for Manuscripts – Use double line-spacing.  Insert continuous line numbering.  ~e submitted text of Research Papers should cover no more than 18 pages (without Tables, Legends, and References, style: font size 12, double line spacing). ~e number of illustrations should not exceed 10. Results may be shown in tables or fgures, but not in both of them.  Please submit, with the manuscript, the names, addres­ses and e-mail addresses of four potential referees. Note that the editor retains the sole right to decide whether or not the suggested reviewers are used. Units and abbreviations Only standard SI symbols and abbreviations should be used in the text, tables and fgures. Symbols for physical quantities in the text should be written in Italics (e.g. v, T, etc.). Symbols for units that consist of letters should INSTRUCTIONS FOR AUTHORS be in plain text (e.g. Pa, m, etc.). All abbreviations should be spelt out in full on frst appearance. Figures Figures must be cited in consecutive numerical order in the text and referred to in both the text and the caption as Fig. 1, Fig. 2, etc. Figures may be saved in any common format, e.g. BMP, JPG, GIF. However, the use of CDR format (CorelDraw) is recommended for graphs and line drawings, since vector images can be easily reduced or enlarged during fnal processing of the paper. When labelling axes, physical quantities (e.g. v, T, etc.) should be used whenever possible. Multi-curve graphs should have individual curves marked with a symbol; the meaning of the symbol should be explained in the fgure caption. Good quality black-and-white photographs or scanned images should be supplied for the illustrations. Tables Tables must be cited in consecutive numerical order in the text and referred to in both the text and the caption as Table 1, Table 2, etc. ~e use of names for quantities in tables should be avoided if possible: correspond­ing symbols are preferred. In addition to the physical quantity, e.g. t (in Italics), units (normal text), should be added on a new line without brackets. Any footnotes should be indicated by the use of the superscript1. LIST OF references Citation in text Please ensure that every reference cited in the text is also present in the reference list (and vice versa). Any refer­ences cited in the abstract must be given in full. Unpub­lished results and personal communications are not recommended in the reference list, but may be mentioned in the text, if necessary. Reference style Text: Indicate references by number(s) in square brack­ets consecutively in line with the text. ~e actual authors can be referred to, but the reference number(s) must always be given: Example: “... as demonstrated [1,2]. Brandl and Blovsky [4] obtained a di˘erent result ...” List: Number the references (numbers in square brackets) in the list in the order in which they appear in the text. Reference to a journal publication: [1] Jeluši°, P., Žlender, B. 2013. Soil-nail wall stability analysis using ANFIS. Acta Geotechnica Slovenica 10(1), 61-73. Reference to a book: [2] Šuklje, L. 1969. Rheological aspects of soil mechan­ics. Wiley-Interscience, London Reference to a chapter in an edited book: [3] Mitchel, J.K. 1992. Characteristics and mechanisms of clay creep and creep rupture, in N. Guven, R.M. Pollastro (eds.), Clay-Water Interface and Its Rheo-logical Implications, CMS Workshop Lectures, Vol. 4, ~e clay minerals Society, USA, pp. 212-244. Conference proceedings: [4] Brandl, H., Blovsky, S. 2005. Slope stabilization with socket walls using the observational method. Proc. Int. conf. on Soil Mechanics and Geotechnical Engineering, Bratislava, pp. 2485-2488. Web references: [5] As a minimum, the full URL should be given and the date when the reference was last accessed. Any further information, if known (DOI, author names, dates, reference to a source publication, etc.), should also be given. Author information ~e following information about the authors should be enclosed with the paper: names, complete postal addresses, telephone and fax numbers and E-mail addresses. Indicate the name of the corresponding author. Acceptance of papers and copyright ~e Editorial Committee of the Slovenian Geotechnical Review reserves the right to decide whether a paper is acceptable for publication, to obtain peer reviews for the submitted papers, and if necessary, to require changes in the content, length or language. On publication, copyright for the paper shall pass to the ACTA GEOTECHNICA SLOVENICA. ~e AGS must be stated as a source in all later publication. For further information contact: Editorial Board ACTA GEOTECHNICA SLOVENICA University of Maribor, Faculty of Civil Engineering, Transportation Engineer­ing and Architecture Smetanova ulica 17, 2000 Maribor, Slovenia E-mail: ags@um.si 48. Acta Geotechnica Slovenica, 2022/2 NAMEN REVIJE Namen revije ACTA GEOTECHNICA SLOVENICA je objavljanje kakovostnih teoreti°nih °lankov z novih pomembnih podro°ij geomehanike in geotehnike, ki bodo dolgoro°no vplivali na temeljne in prakti°ne vidike teh podro°ij. ACTA GEOTECHNICA SLOVENICA objavlja °lanke s podro°ij: mehanika zemljin in kamnin, inženirska geologija, okoljska geotehnika, geosintetika, geotehni°ne konstrukcije, numeri°ne in analiti°ne metode, ra°unal­niško modeliranje, optimizacija geotehni°nih konstruk­cij, terenske in laboratorijske preiskave. Revija redno izhaja dvakrat letno. AVTORSKE PRAVICE Ko uredništvo prejme °lanek v objavo, prosi avtorja(je), da prenese(jo) avtorske pravice za °lanek na izdajatelja, da bi zagotovili kar se da obsežno razširjanje informacij. Naša revija in posamezni prispevki so zaš°iteni z avtorskimi pravicami izdajatelja in zanje veljajo naslednji pogoji: Fotokopiranje V skladu z našimi zakoni o zaš°iti avtorskih pravic je dovoljeno narediti eno kopijo posameznega °lanka za osebno uporabo. Za naslednje fotokopije, vklju°no z ve°kratnim fotokopiranjem, sistemati°nim fotoko­piranjem, kopiranjem za reklamne ali predstavitvene namene, nadaljnjo prodajo in vsemi oblikami nedobi°k­onosne uporabe je treba pridobiti dovoljenje izdajatelja in pla°ati dolo°en znesek. Naro°niki revije smejo kopirati kazalo z vsebino revije ali pripraviti seznam °lankov z izvle°ki za rabo v svojih ustanovah. Elektronsko shranjevanje Za elektronsko shranjevanje vsakršnega gradiva iz revije, vklju°no z vsemi °lanki ali deli °lanka, je potrebno dovoljenje izdajatelja. ODGOVORNOST Revija ne prevzame nobene odgovornosti za poškodbe in/ali škodo na osebah in na lastnini na podlagi odgo­vornosti za izdelke, zaradi malomarnosti ali druga°e, ali zaradi uporabe kakršnekoli metode, izdelka, navodil ali zamisli, ki so opisani v njej. AIMS AND SCOPE ACTA GEOTECHNICA SLOVENICA aims to play an important role in publishing high-quality, theoretical papers from important and emerging areas that will have a lasting impact on fundamental and practical aspects of geomechanics and geotechnical engineering. ACTA GEOTECHNICA SLOVENICA publishes papers from the following areas: soil and rock mechan­ics, engineering geology, environmental geotechnics, geosynthetic, geotechnical structures, numerical and analytical methods, computer modelling, optimization of geotechnical structures, feld and laboratory testing. ~e journal is published twice a year. COPYRIGHT Upon acceptance of an article by the Editorial Board, the author(s) will be asked to transfer copyright for the article to the publisher. ~is transfer will ensure the widest possible dissemination of information. ~is review and the individual contributions contained in it are protected by publisher’s copyright, and the following terms and conditions apply to their use: Photocopying Single photocopies of single articles may be made for personal use, as allowed by national copyright laws. Permission of the publisher and payment of a fee are required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Subscribers may reproduce tables of contents or prepare lists of papers, including abstracts for internal circula­tion, within their institutions. Electronic Storage Permission of the publisher is required to store electron­ically any material contained in this review, including any paper or part of the paper. RESPONSIBILITY No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of product liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein.