Procter & Gamble (United Kingdom)
GraFar - Repository of the Faculty of Civil EngineeringNot a member yet
3934 research outputs found
Sort by
Experimental investigation of the construction joint in concrete ground floors
Get PDFGround floors of industrial buildings, fully supported on the ground or piles, are usually made
of several concrete segments connected through free-movement construction joints. These joints
should provide shear load transfer of the slab and minimise its vertical displacements while
allowing slab horizontal movement to prevent damage due to concrete dry shrinkage. The
construction joint “Dilat 08”, consisting of round steel dowels and steel formwork, was
experimentally tested in a laboratory environment to determine its behaviour to shear load. The
experimental double-shear setup consisted of three slabs connecting through the “Dilat 08” join
system, with two side slabs fully supported on the base and the middle slab on which the vertical
loading was applied. Specimens were formed with a gap of 10 mm between adjacent slabs to
simulate the construction joint opening. Two construction joint orientations were investigated
introducing the vertical load to the slab with a steel plate on the edge and to the slab with an
angle profile on the edge. Experimental testing of both specimens resulted in construction joint
failure due to concrete cracking around steel dowels, near the edges of the middle slab. The
investigation confirmed that the joint orientation affects the joint shear response, demonstrating
larger joint resistance when the steel plate is installed in the slab subjected to failure. Although
contributing to the joint bending stiffness before formwork installation on the site, the horizontal
leg of the angle profile decreases the joint resistance. Experimentally obtained joint resistance
was compared to the design values of resistance according to Technical Report 34 and EN 1992-
4, highlighting a considerable difference between the two analytical approaches. The comparison
between the “Dilat 08” and three types of free-movement joints available on the construction
market showed that “Dilat 08” features larger resistance according to Technical Report 34, but
it also weighs more
Numerical modelling of the seismic response of confined masonry walls
No full textTehnologija izgradnje zidova sa AB serklažima u zidanim konstrukcijama pruža ubedljivo
rešenje za izgradnju stambenih zidanih zgrada u regionima podložnim seizmičkoj opasnosti sa
umerenim do visokim seizmičkim rizikom. Svrha ovog rada je da se ispita seizmička otpornost
novorazvijene tehnologije modernih zidova sa AB serklažima građenih od glinenih blokova sa
velikim šupljinama za termoizolaciju, koji se lepe poliuretanskim lepkom umesto malterom. Na
osnovu rezultata dva eksperimentalna ciklična testa smicanja razvijen je detaljan numerički
model zasnovan na metodi konačnih elemenata u softveru Abaqus. Razvijeni 3D mikro model
je bio uspešan jer je bio sposoban da precizno predvidi globalni odgovor zidova u smislu krive
sila-pomeranje (pushover krive), uzorka oštećenja u zidu i AB serklažima, kao i lokalnog efekta
„odsecanja“ isturenog dela zida
Preparing for revitalization of the turbines of the HE Đerdap 2: Measurement of velocity distribution at turbine inlets
No full textKontinualno merenje protoka na brojnim turbinama hidroelektrana se standardno radi nekom od relativnih metoda (na primer Winter-Kennedy). Umesto da se meri složeno polje brzina u jednoj proticajnoj ravni i da se njenom integracijom dobije protok, meri se samo jedna karakteristična veličina a na fizičkom modelu se odrede parametri preslikavanja vrednosti te veličine u trenutni protok. Najčešće se koristi Winter-Kennedy metoda gde se merenjem razlike pritisaka dobija relativna (indeksna) vrednost protoka. Merna nesigurnost tako određenog protoka je znatno veća od nesigurnosti merenja ostalih relevantnih veličina za određivanje optimalnih radnih uslova turbine. Da bi se smanjila merna nesigurnost, potrebno je „apsolutnim merenjima“ celog polja brzina odrediti trenutni protok i preračunati, u realnim uslovima, korekcije indeksne metode. To se posebno odnosi na hidroelektrane čija dispozicija nije „idealna“ kao što je bila na fizičkom modelu, kao što je HE „Đerdap 2“ sa poznatim problemom „kosog dostrujavanja“. Sa ciljem bolje procene hidrauličke efikasnosti turbina i prikupljanja podataka o realnim uslovima rada turbina i ulazne rešetke, a zbog planiranih radova na revitalizaciji, projektovan je i primenjen inovativni sistem za apsolutno merenje protoka koji je prikazan u ovom radu. Na pokretni ram, pozicioniran na ulazu u turbinu uzvodno od grube rešetke, postavljeno je 15 elektromagnetnih (EM) senzora u jednoj horizontalnoj ravni, zajedno sa dva redudantna akustična Doppler senzora. Svaki od senzora meri sve tri komponente brzina. Ram se podiže duž cele visine proticajnog preseka snimajući celo polje brzina. Položaj rama se prati pomoću dva enkodera, dok se dva senzora pritiska koriste za merenje dubine vode. Merenja su sinhronizovana sa lokalnim SCADA sistemom odakle se preuzimaju podaci o radu turbine. Uvažavajući specifičnosti mernog sistema, novorazvijenih EM sondi i postojećih hidrauličkih uslova, razvijena je adekvatna procedura za procenu nesigurnosti izmerenog protoka. U ovom radu je prikazana merna metoda i dati su neki rezultati merenja na agregatima HE „Đerdap 2“.The continuous flow measurement on hydroelectric turbine units is conventionally carried out using the relative methods (mostly Winter-Kennedy), where the complex velocity field in flow cross section is replaced by a single measured quantity. On a physical model, mapping parameters are determined. The Winter-Kennedy method is most used, where this relative (index) flow quantity is obtained by measuring differential pressure. The measurement uncertainty of the determined flow using this method is usually significantly higher than the uncertainty of other relevant quantities for determining the optimal turbine operating conditions. To reduce the flow uncertainty, it is necessary to perform "absolute measurements" of the entire velocity field and determine the necessary corrections to the index method under real operating conditions. This is especially true for plants where disposition is different from one used on physical model, which is the case of “Đerdap 2” and recognized problem of skewed inflow. With the aim of better assessing the hydraulic efficiency of turbines and to collect data on the actual operating velocity distribution in the inlet section in front of trash-rack needed for planned revitalization works, an innovative system for absolute flow measurement has been designed and implemented. The moving frame was designed and installed at the turbine inlet, upstream from the trash-rack. The 15 electromagnetic (EM) sensors and two redundant acoustic Doppler sensors were installed on horizontal bar. Each sensor measures all three velocity components. The frame moves by lifting along the entire height of the flow section, allowing the entire velocity field to be scanned. The position of the frame is tracked by two encoders, while two pressure sensors are used to measure the water depth. Measurements are synchronized with the local SCADA system, so appropriate turbine operation data are also used. Considering the specificities of the measurement system itself with newly developed EM sensors, as well as hydraulic conditions, an adequate procedure has been developed to assess the uncertainty of the measured flow. This paper presents the measurement method and provides some measurement results on the units of the “Đerdap 2” hydroelectric power plant
Innovative solution for decoupling RC frames from masonry infills with openings: experimental results
No full textMasonry infills are a common solution for inner and outer partitions in RC frame structures. However, uncontrolled and heavy damage to masonry infills in recent earthquake events often caused by the interaction of masonry infills with RC frames showed the high seismic vulnerability. Severe damage to masonry infills results in high economic losses and presents a threat to human safety. The bad seismic performance of masonry infills is a strong motivation for finding solutions that can prevent or reduce the damage to masonry infills. This paper presents the principles of an innovative system for decoupling of masonry infills from the surrounding frame. The decoupling is achieved by inserting recycled rubber strips between masonry infills and RC frame. The installed decoupling system postpones the activation of masonry infills under in-plane seismic loads and provides support under out-of-plane seismic loads. Experimental results on RC frames with rigidly connected infills and RC frames with decoupled infills are presented and compared for infills with and without openings. Results of the study show the highly improved seismic performance of masonry infills with the innovative decoupling system
Effect of rib-to-beam angle on the shear resistance of headed studs in composite slab
No full textHeaded stud shear connectors are commonly implemented in steel-concrete composite floors for connecting the concrete slab to the steel beam and providing composite action between those two elements. The orientation of profiled sheeting is usually transverse or parallel to the supporting beam, and for those cases, design codes provide specific detailing requests and relations for obtaining headed stud resistance. Nevertheless, design codes do not recognise specific configurations when the angle between profiled sheeting ribs and the beam is in the range between 0° and 90°, which is characteristic of irregular-shaped floor structures. Due to the lack of design recommendations for headed stud resistance in these cases, investigations of the effect of the rib-to-beam angle on headed stud shear performance were conducted and presented in this paper. Experimental work included push-out tests conducted on specimens with rib-to-beam angles of 45° and 60°, as well as control specimens with ribs transverse to the beam. The influence of varied rib-to-beam angles on shear connector properties such as shear resistance, stiffness, ductility, and failure mode was discussed. According to experimental research, finite element models of push-out tests were created and verified. The parametric numerical analysis covered key parameters such as concrete class, headed stud height and diameter, and rib-to-beam angle. According to the results, an analytical relation for obtaining headed stud resistance in profiled steel sheeting with a rib-to-beam angle between 30° and 90° was proposed
Primena bio-veziva za asfaltne slojeve kolovoznih konstrukcija
No full textBitumen, kao osnovno vezivo koje se primenjuje za izradu asfaltnih slojeva kolovoznih konstrukcija, se dobija destilacijom sirove nafte, koja će kao fosilno gorivo, imati ograničenu dostupnost u budućnosti i čija eksploatacija ima značajan negativan uticaj na okolinu. Koncept bio-veziva je nedavno razvijen i odnosi se na veziva za asfaltne mešavine koja sadrže bio-obnovlјive materijale. U okviru rada su prikazana tipovi bio-veziva, njihova primena, osnovni zahtevi da bi mogli da budu primenjena u asfaltnim slojevima, kao i postupci proizvodnje. Prikazane su osnovne fizičke i hemijske karakteristike bio-ulja, kao i značajne karakteristike bio-veziva i njihov uticaj na otpornost na starenje modifikovanih bitumena. Konačno, dat je prikaz uticaja primene bio-veziva na ponašanje asfaltnih mešavina, koji u znatnoj meri zavisi od porekla biomase korišćene za dobijanje bio-veziva. Bio-veziva generalno utiču povoljno na poboljšanje karakteristika asfaltnih mešavina na niskim temperaturama, ali dovode do smanjenja otpornosti mešavina na kolotrage na visokim temperaturama. Pored ovoga, dat je i prikaz primene bio veziva za regeneraciju ostarelog bitumena iz struganog asfalta koji se u velikoj meri primenjuje prilikom recikliranja asfaltnih mešavina
Transformers and Tokenization: A Novel Approach to Predicting In-game Purchases
No full textPlayer purchases in free-to-play (F2P) online games represent essential metrics for assessing
user engagement and understanding behavioral patterns. Predictive models for player purchases empower game developers to tailor marketing strategies, efficiently allocate resources, and enhance player retention through data-driven insights. This proactive approach maximizes revenue and fosters player satisfaction, providing a competitive edge in the dynamic gaming industry.
This work proposes a novel methodology for player purchase prediction, leveraging Transformer neural networks in combination with a discretized representation of player behavioral histories. The methodology assumes that recent player activity plays a pivotal role in shaping future behavior while acknowledging that similar patterns may yield different outcomes depending on the broader context of the player’s cumulative history. By discretizing behavioral features and utilizing tokenized inputs derived from this history, the approach predicts whether a player will make a purchase within the next 3, 5, or 7 days. The dataset, derived from the mobile game Woka Woka by Two Desperados (https://twodesperados.com), captures comprehensive daily player activity, including engagement metrics (e.g., sessions, time played), skill-related measures (e.g., levels completed), and purchase behavior (e.g., in-game currency transactions and balances). These data enable a focused analysis of player engagement, skill, and purchasing
tendencies while excluding features related to player interaction due to the game's single-player nature. Each player's daily data is encoded into a sequence of 13 tokens, corresponding to 13 features derived from the available dataset. The first two tokens represent the player’s presence in the game and whether a transaction occurred on that day. For the remaining features, four tokens are used depending on the feature's value, capturing key behavioral and transactional information.
The performance of the proposed approach was validated through comparisons with standard machine learning models, including Random Forest, XGBoost, and Multilayer Perceptron. Given that these models require fixed-length inputs, player histories were represented as fixed-length vectors. Two alternative representations were evaluated: one based on the frequency of each token in the player's history and another using statistical features (minimum, maximum, first quartile, median, and third quartile) calculated for each original feature before tokenization.
Model performance was assessed using the macro-averaged F1 score, a widely recognized metric for multiclass classification. To address the challenge of class imbalances, additional metrics were employed, including the Area Under the Receiver Operating Characteristic Curve (AUROC) and the Matthews Correlation Coefficient (MCC). The AUROC evaluates the model's ability to distinguish between classes, while the MCC offers a balanced measure of performance, accounting for true and false positives and negatives across all classes. The results demonstrate the exceptional performance of the Transformer Neural Network model, particularly when leveraging the Self-Attention mechanism. This model outperformed traditional approaches, such as Random Forest, XGBoost, and Multilayer Perceptron, across all prediction periods. Furthermore, models incorporating feature statistics slightly surpassed those using token frequency, as tokenization inherently approximates feature value distributions, leading to some loss of information. Nevertheless, the Transformer Neural Network consistently delivered superior results, underscoring the importance of capturing long-term player behavior patterns to enhance predictive accuracy. In conclusion, while discretization introduces a degree of information loss, the Transformer model effectively compensates by capturing nuanced, long-term patterns in player behavior. This study provides a significant contribution to the field of in-game purchase prediction, demonstrating the potential of advanced neural network architectures to outperform traditional methods. The findings pave the way for further innovation in predictive analytics within the gaming industry
Adaptability of flood embankments to climate change
No full textIn the context of growing concerns about climate change affecting precipitation patterns, there is a need to incorporate these changes into the design of flood defenses, particularly earthfill embankments. Although numerous geotechnical modeling techniques have improved understanding of soil embankment behavior, uncertainty regarding the impact of climate change to embankment performance still remains. Research in this field aims to identify instabilities that potentially could occur throughout an embankment’s lifecycle, and development of corresponding solutions. Geotechnical modeling techniques and site-specific precipitation pattern projections, establish a comprehensive analysis framework, which allows assessing long-term behavior and resilience to a changing climate of existing and new embakments. The paper provides an overview of the climate-driven processes that are of greatest concern for hydro-mechanical behavior of earthfill flood embankments. Then, key aspects that need to be considered for predictive computational modeling, which combines geotechnical and climate modeling, are presented. Finaly, various strategies for embankment adaptation in order to improve their resilience to climate change are given
Business analysis of the leading construction companies in Serbia
No full textAnalysis of the construction company's operations is a source of very useful data that is used for planning the company's work. The data from this analysis is used for comparison with other, similar companies on the market and as a measure of business success. The purpose of the research presented in this paper is to present the state of business of the leading construction companies involved in road construction in Serbia. In order to perform an adequate analysis, the parameters that will be observed are defined. The expected values of these parameters for construction companies are shown. The key performance indicators (KPI) presented in the publicly published documents of the companies were analyzed. The paper analyzed and compared data for 10 companies with the highest revenue during 2023. Also, the trend of changes in some of the parameters in the period from 2020 to 2023 is shown. The results presented in the paper can be indicators of the state of the market in Serbia and serve as a measure to control the work of similar companies
Invertibility of some special matrices involving certain numbers
Get PDFA family of structured matrices, such as Toeplitz, Hankel, Vandermonde, circulant, Hessenberg, Cauchy, and other well-known families of special matrices, offering extensive utility in digital signal processing, image manipulation, coding theory, statistics, quantum mechanics, and more. A particularly intriguing scenario arises when their elements are derived from k−Horadam numbers or Chebyshev polynomials, encompassing Fibonacci, Lucas, and Pell numbers. Additionally, the case where matrix entries are composed of generalized Narayana numbers is also of interest. We will present the conditions under which circulant and skew circulant matrices involving generalized Narayana numbers are invertible. These findings not only have practical applications across various fields but also contribute to advancing our comprehension of structured matrices and their
inherent characteristics