Universidad de Concepción Facultad de Ingeniería

Laboratorio de Gestión de Activos

Publicaciones | Publicaciones Descargables

Año Referencia Revista
2019 Reducing mining footprint by matching haul fleet demand and route-oriented tire types. R. Pascual, M. Román, M. López-Campos, M. Hitch, E. Rodovalho.
Off-The-Road (OTR) tires represent an important part of the operational costs of the mining industry. Each year, a typical operation consumes hundreds of tires. In terms of operational costs, tires are only second to fuel, translating into several USD millions per year for an average mine operation. In addition, tires affect equipment performance and availability and, as a consequence, put at risk the capacity of the haul fleets to deliver the production targets. OTR tire lifespan depends on proper type selection. Each tire-type implies choosing a combination of rubber compounds and geometric specifications that are suited to road parameters. Medium and long term mining plans specify routes and production goals. In general, each route has a specific optimal tire type. The traditional approach is to consider the most demanding conditions and selecting a single tire type for the whole fleet . In such a way, truck dispatch is flexible as any truck can haul on any route. A drawback is that this one-size-fits-all policy increases tire consumption as the worst-case route sets the type of tire for the entire fleet. The above builds an interesting case for optimizing tire selection and haul fleet usage schedule as both decisions can be relevant in search for reducing tire consumption, decrease operational costs, and assure production plan adherence. As tire type may influence haul cycle times, assignment of trucks to different routes should also be considered. This work introduces a novel methodology for setting a usage allocation plan for the haul fleet and selecting route-oriented tire types. We test our methodology using a medium size open-pit operation in northern Chile. The case study shows a tire consumption reduction of 7.3% with respect to the traditional approach over a 5 year time span. Net present tire costs are reduced by USD 2.7 millions (-7.2%). Our methodology presents a novel approach to both reducing costs and achieving long-term production plans.
Journal of Cleaner production
2019 A Novel Deep Capsule Neural Network for Remaining Useful Life Estimation. Andrés Ruiz-Tagle Palazuelos, Enrique López Droguett1 and Rodrigo Pascual.
With the availability of cheaper multisensor systems, one has access to massive and multi-dimensional sensor data for fault diagnostics and prognostics. However, from a time, engineering and computational perspective, it is often cost prohibitive to manually extract useful features and to label all the data. To address this challenges, deep learning techniques have been used in the recent years. Within these, convolutional neural networks (CNN) have shown remarkable performance in fault diagnostics and prognostics. However, CNNs do present limitations from a prognostics and health management perspective: to improve its feature extraction generalization capabilities and reduce computation time, ill-based pooling operations are employed, which require sub-sampling of the data, thus loosing potentially valuable information regarding an asset’s degradation process. Capsule Neural Networks (CapsNets) have been recently proposed to address these problems with strong results in computer vision related classification tasks. This has motivated us to extend CapsNets for fault prognostics and, in particular, remaining useful life estimation. The proposed model, architecture and algorithm are tested and compared to other state-of-the art deep learning models on the benchmark C-MAPPS turbofans dataset. The results indicate that the proposed CapsNets are a promising approach for RUL prognosticsfrom multi-dimensional sensor data.
Journal of Risk and Reliability
2018 An asset-management oriented methodology for mine haul-fleet usage scheduling. C. Nakousi, R. Pascual, A. Anani, F. Kristjanpoller, P. Lillo.
Different complexities force mining companies to find efficient ways to respond to demand challenges and ensure long-term sustainability. It explains, in part, the increase in the use of prescriptive analytics to optimize physical-asset life-cycle costs and decrease greenhouse gas (GHG) emissions. Mining, being an asset-intensive industry, provides huge improvement opportunities. This is especially true for scheduling practices of mine haulage fleet usage in long term planning. Fleet aging implies important cost increases in maintenance and repair (M&R), and overhauls. Fleets are often heterogeneous in term of truck performance, fuel consumption and GHG emissions. Sub-optimal scheduling decisions may induce severe cost over-runs and increased emissions. This paper proposes an original mixed integer programming formulation to optimize mine haulage equipment scheduling in the long term. The model considers the effects of equipment aging, fuel consumption, payload capacity and cycle times. Our formulation handles different aspects that according to author's knowledge have not been considered in the literature as a whole: (i) joint minimization of fuel, M&R, and overhaul costs, (ii) reduction of GHG emissions, (iii) heterogeneous equipment performance metrics, (iv)increase in cycle times due to mine aging. The case study shows a cost reduction of 13% in the discounted .ows associated with fuel, M&R, and overhauls in a time horizon of 10 years. This .gure translates into an NPV gain of 13.1 million USD. Additionally, GHG emissions are reduced by an average of 3,470 tons/year or 11% overall.
Reliability Engineering and System Safety
2018 Corrigendum to "Reliability Assessment Methodology for Massive Manufacturing Using Multi-Function Equipment". M. López-Campos, F. Kristjanpoller, P. Viveros, and R. Pascual.
-
Complexity
2018 A special issue on Statistical methods in mining industry. Orietta Nicolis and Rodrigo Pascual.
In this special issue, we selected four papers where some statistical methods are used for taking decisions, assessing the security of mining equipments, and predicting the proportion of a certain mineral in rock samples. While the papers of Neira et al and Stehlík et al use the concepts of the reliability theory based on typical failure distributions (such as the Weibull, Gamma, and Beta distributions), the work of Jiang et al uses nonparametric statistical methods for analyzing the failures. A different issue has been proposed by Huerta et al for predicting the proportion of a mineral in rocks. In particular, in the work of Neira et al, a statistical method inspired from reliability engineering has been proposed for monitoring the “health” of microseismic sensing networks, that is, its capability to properly register all the microseismic activity above a certain energy level. The monitoring of the health of a microseismic network has been addressed in this work by describing and characterizing the faulty behavior of each sensor in analogy with standard ideas and methods of reliability theory. In particular, the proposed method analyzes two relevant features of each of the sensors' behavior, namely, what type of noise is or might be affecting the registering process and how effective at registering microseisms the sensor is. Once the noisy activations and the types of noise have been identified, a Weibull distribution has been fitted to the time differences of consecutive noisy activations (failures). The estimated parameters are then used to compute the reliability of each sensor with respect to the different types of noise. In order to estimate the noise that is present in the seismograms, three indicators have been proposed: two of these are based on the power spectral density (PSD) and the third is given by the signal-to-noise ratio (SNR). In all cases, the computed value of the indicator is compared with a threshold to assess if an activation is noisy or not. Then, the ratio of activations is extracted by evaluating the quotient between the number of satisfactory activations registered by a sensor and the number of satisfactory activations that the sensor was supposed to register in a given period of time. Also, the authors propose a new indicator, computed for each individual sensor, that conveys both the information of the reliability of the sensors and their ratio of activations. The proposed methodology is then applied to the microseismic data registered at the Chilean underground mine El Teniente. The study illustrates the capability of the proposed methodology to discriminate and rank sensors with satisfactory, poor, or defective sensing performances, as well as to characterize their failure profile or type of information that can be used to plan or optimize the network maintenance procedures.
Applied Stochastic Modelling in Business and Industry
Diseño Web - Posicionamiento Web