Search results

Construction plant breakdown affects projects by prolonging duration and increasing costs. Therefore, prediction of plant breakdown, as a precursor to conducting timely maintenance works, cannot be underestimated. This paper thus sought to develop a model for predicting plant breakdown time from a sequence of discrete plant breakdown measurements that follow non‐random orders. An ARIMA (1,1,0) model was constructed following experimentation with exponential smoothening. The model utilised breakdown observations obtained from six wheeled loaders that had operated a total of 14,467 hours spread over a 300‐week period. The performance statistics revealed MAD and RMSE of 5.03 and 5.33 percent respectively illustrating that the derived time series model is accurate in modelling the dependent variable. Also, the F‐statistics from the ANOVA showed that the type and frequency of fault occurrence as a predictor variable is significant on the model's performance at the five percent level. Future work seeks to consider a more in depth multivariate time series analyses and compare/contrast the results of such against other deterministic modelling techniques.

Effective management and utilisation of plant history data can considerably improve plant and equipment performance. This rationale underpins statistical and mathematical models for exploiting plant management data more efficiently, but industry has been slow to adopt these models. Reasons proffered for this include: a perception of models being too complex and time consuming; and an inability of their being able to account for dynamism inherent within data sets. To help address this situation, this research developed and tested a web‐based data capture and information management system. Specifically, the system represents integration of a web‐enabled relational database management system (RDBMS) with a model base management system (MBMS). The RDBMS captures historical data from geographically dispersed plant sites, while the MBMS hosts a set of (Autoregressive Integrated Moving Average – ARIMA) time series models to predict plant breakdown. Using a sample of plant history file data, the system and ARIMA predictive capacity were tested. As a measure of model error, the Mean Absolute Deviation (MAD) ranged between 5.34 and 11.07 per cent for the plant items used in the test. The Root Mean Square Error (RMSE) values also showed similar trends, with the prediction model yielding the highest value of 29.79 per cent. The paper concludes with direction for future work, which includes refining the Graphical User Interface (GUI) and developing a Knowledge Based Management System (KBMS) to interface with the RDBMS.

Hiring, selecting or assessing plant operatives' proficiency in the UK construction industry is an increasingly difficult task. A number of plant operator certification schemes are available to practitioners and each scheme trains to a myriad of bespoke standards. Consequently, the decision to employ a candidate often rests upon the employer's intuition and judgement and creates an unnecessary dilemma. To address this aforementioned problem, findings of research work that modelled plant operators' maintenance proficiency is presented. A UK nationwide survey was conducted to elicit plant professional opinion on what ‘training and educational’ (T&E) attributes constitute ‘good’ operator proficiency. The data was then arranged into three categories of operator maintenance proficiency: good, average and poor Multivariate Discriminant Analysis (MDA) was used on 75 percent of a simulated data set. The model utilised five T&E attributes, namely: duration of training provided, operator holder of alternative training card (not Certificate of Training Achievement (CTA) or Scottish/National Vocational Qualifications (S/NVQ)), operator's oral communication skills, operator's planning skills and operator's mechanical knowledge. Performance analysis revealed that model classification accuracy was 89.10 percent. The remaining 25 percent hold out sample was then modelled for validation purposes using the derived MDA model. Accuracy of the sub‐sample model was high at 77.60 percent whilst a paired sample T‐tests for the 75 percent and 25 percent sample data established that there was no significant statistical difference between actual and predicted classifications. Future work is proposed that aims to model other factors that influence operator maintenance proficiency; namely, work situational, motivational management and personal factors.

The productivity and output levels of construction plant and equipment depends in part upon a plant operator’s maintenance proficiency; such that a higher degree of proficiency helps ensure that machinery is maintained in good operational order. In the absence of maintenance proficiency, the potential for machine breakdown (and hence lower productivity) is greater. Using data gathered from plant and equipment experts within the UK, plant operators’ maintenance proficiency are modelled using a radial basis function (RBF) artificial neural network (ANN). Results indicate that the developed ANN model was able to classify proficiency at 89 per cent accuracy using 10 significant variables. These variables were: working nightshifts, new mechanical innovations, extreme weather conditions, planning skills, operator finger dexterity, years experience with a plant item, working with managers with less knowledge of plant/equipment, operator training by apprenticeship, working under pressure of time and duration of training period. It is proffered that these variables may be used as a basis for categorizing plant operators in terms of maintenance proficiency and, that their potential for influencing operator training programmes needs to be considered.

Utilization of off‐highway vehicles forms an essential part of UK industry's efforts to augment the productivity of plant operations and reduce production costs. However, uninterrupted utilization of plant and equipment is requisite to reaping the maximum benefit of mechanization; one particular problem being plant breakdown duration and its impact upon process productivity. Predicting the duration of plant downtime would enable plant managers to develop suitable contingency plans to reduce the impact of downtime. This paper presents a stochastic mathematical modelling methodology (more specifically, probability density function of random numbers) which predicts the probable magnitude of ‘the next’ breakdown, in terms of duration for tracked hydraulic excavators. A random sample of 33 machines was obtained from opencast mining contractors, containing 1070 observations of machine breakdown duration. Utilization of the random numbers technique will engender improved maintenance practice by providing a practical methodology for planning, scheduling and controlling future plant resource requirements. The paper concludes with direction for future research which aims to: extend the model's application to cover other industrial settings and plant items; to predict the time at which breakdown will occur (vis‐à‐vis the duration of breakdown); and apply the random numbers modelling to individual machine compartments.

Hydraulic excavator cycle time and associated unit costs of excavation for given input estimating data, for machines operating in the UK construction industry, are predicted. Using multiple regression analysis, three variables are identified as accurate predictors of cycle time: machine weight, digging depth and machine swing angle. With a coefficient of determination (R2) of 0.88, a mean percentage error (MPE) of −5.49, and a mean absolute error (MAPE) of 3.67, the cycle time model is robust; this is further validated using chi‐square analysis and Pearson's correlation coefficient (on predicted and actual values of machine cycle time). An illustrative example of the model's application to determine machine productivity is given. The paper concludes with a spreadsheet model for calculating excavation costs (m3 and cost per h) which is able to deal with any combination of the three independent cycle time predictor variables and other estimator's input data.

Much rhetoric exists on the urgency of transitioning from current practices to a more sustainable society. However, because this imperative is guided by strong ideological overtones, weaknesses and failures in the transition effort attract inadequate scrutiny. This paper reviews Australia's progress with sustainability in an urban domain and identifies key issues hindering the sustainability transition effort.

Research on urban sustainability is ubiquitous but this weight of publications tends to emphasize technical, operational or prescriptive themes. This research uses an interpretivist philosophical lens and inductive reasoning to manually analyse pertinent literature sourced from the Scopus and Web of Science data-bases. Specifically, this study assembles outcome and evaluative assessments pertaining to Australia's urban sustainability efforts to identify both the progress achieved and residual structural impediments.

Emergent findings illustrate that Australia's urban sustainability goals, as expressed by the Paris Accord, have not been met. Obstruction can be attributed to over-ambitious objectives combined with weak federal leadership, under-resourced local government, over-reliance on superficial rating systems and an ineffective regulatory regime. Elite “green branding” by image conscious corporations are insufficient to offset the general disinterest of the unincentivized majority of building owners and developers.

This paper cogently summarizes Australia's urban sustainability status, along with complexity of the challenges it faces to meet targets set.

The Control of Vibration at Work Regulations (CVWR), quantify workplace vibration exposure using exposure action, and exposure limit values (EAV and ELV respectively). Hand‐arm vibration (HAV) risk can be objectively assessed using hand‐tool vibration magnitude data, for comparison to the EAV and ELV. When considering risk controls, one disadvantage of this ‘focus’ on vibration magnitude, is that it might deflect appreciation of the economic implications of such controls, resulting from for example: restrictions on tool usage time; the need for operator rotas where continuous tool use is required; and complications in estimating labour costs because of these types of condition. Based on a sample of hand‐tools’ performance data, this research developed ‘hybrid’ (performance/vibration) dimensions for quantifying tools’ efficacy; representing (interalia) units of work achievable to reach the EAV and ELV. These hybrid dimensions characterize an alternative performance‐based (and therefore financially related) way of considering a tool’s ‘suitability’ within CVWR parameters; over and above the (selection) criterion of tool vibration magnitude. Analyses are then presented that investigate the time and cost ramifications of using multiple operators, to sustain continuous tool usage while keeping exposure levels within CVWR limits.

Safety research has focused on drivers, pedestrians and vehicles, with scarce attention given to highway traffic officers (HTOs). This paper develops a robust prediction model which enables highway safety authorities to predict exclusive incidents occurring on the highway such as incursions and environmental hazards, respond effectively to diverse safety risk incident scenarios and aid in timely safety precautions to minimise HTO incidents.

Using data from a highway incident database, a supervised machine learning method that employs three algorithms [namely Support Vector Machine (SVM), Random Forests (RF) and Naïve Bayes (NB)] was applied, and their performances were comparatively analysed. Three data balancing algorithms were also applied to handle the class imbalance challenge. A five-phase sequential method, which includes (1) data collection, (2) data pre-processing, (3) model selection, (4) data balancing and (5) model evaluation, was implemented.

The findings indicate that SVM with a polynomial kernel combined with the Synthetic Minority Over-sampling Technique (SMOTE) algorithm is the best model to predict the various incidents, and the Random Under-sampling (RU) algorithm was the most inefficient in improving model accuracy. Weather/visibility, age range and location were the most significant factors in predicting highway incidents.

This is the first study to develop a prediction model for HTOs and utilise an incident database solely dedicated to HTOs to forecast various incident outcomes in highway operations. The prediction model will provide evidence-based information to safety officers to train HTOs on impending risks predicted by the model thereby equipping workers with resilient shocks such as awareness, anticipation and flexibility.

Conducts an in‐depth examination of the current Certification of Training Achievement (CTA) scheme and critically appraises the role of construction plant operatives within the UK construction industry. Reveals a cacophony of practitioner disapproval of the CTA scheme and the Intermediate Construction Certificate (ICC) route towards attaining the National Vocational Qualification/Scottish Vocational Qualification (NVQ/SVQ) standard.