Continuous provision of quality supply air to data center’s IT pod room is a key parameter in ensuring effective data center operation without any down time. Due to number of possible operating conditions and non-linear relations between operating parameters make the working mechanism of data center difficult to optimize energy use. At present industries are using computational fluid dynamics (CFD) to simulate thermal behaviour for all types of operating conditions. The focus of this study is to predict Supply Air Temperature using Artificial Neural Network (ANN) which can overcome limitations of CFD such as high cost, need of an expertise and large computation time. For developing ANN, input parameters, number of neurons and hidden layers, activation function and the period of training data set were studied. A commercial CFD software package 6sigma room is used to develop a modular data center consisting of an IT pod room and an air-handling unit. CFD analysis is carried out for different outside air conditions. Historical weather data of 1 year was considered as an input for CFD analysis. The ANN model is “trained” using data generated from these CFD results. The predictions of ANN model and the results of CFD analysis for a set of example scenarios were compared to measure the agreement between the two. The results show that the prediction of ANN model is much faster than full computational fluid dynamics simulations with good prediction accuracy. This demonstrates that ANN is an effective way for predicting the performance of an air handling unit.

The use of industrial robots in modern manufacturing scenarios is a rising trend in the engineering industry. Currently, industrial robots are able to perform pre-programmed tasks very efficiently irrespective of time and complexity. However, often robots encounter unknown scenarios and to solve those, they need to cooperate with humans, leading to unnecessary downtime of the machine and the need for human intervention. The main aim of this study is to propose a method to develop adaptive industrial robots using Machine Learning (ML)/Machine Vision (MV) tools. The proposed method aims to reduce the effort of re-programming and enable self-learning in industrial robots. The elaborated online programming method can lead to fully automated industrial robotic cells in accordance with the human-robot collaboration standard and provide multiple usage options of this approach in the manufacturing industry. Machine Vision (MV) tools used for online programming allow industrial robots to make autonomous decisions during sorting or assembling operations based on the color and/or shape of the test object. The test setup consisted of an industrial robot cell, cameras and LIDAR connected to MATLAB through a Robot Operation System (ROS). The online programming tests and simulations were performed using Virtual/Augmented Reality (VR/AR) toolkits together with a Digital Twin (DT) concept, to test the industrial robot program on a digital object before executing it on the real object, thus creating a safe and secure test environment.

Metal Big Area Additive Manufacturing (mBAAM) is a promising approach to large-scale metal additive manufacturing (AM) or 3D printing. The mBAAM system uses an arc-based wire-fed welding robot to build metal parts. A multi-degree-of-freedom robotic arm is known for its extensive range of motion and reliable tool handling. Attaching a torch end-effector to a robotic arm gives it welding capabilities; however, this decreases the motion range and dynamics of the robot. As a result, build volume and printing accuracy are decreased. Additionally, only a portion of time is spent printing in an arc-based process. Maintenance leads to downtime on the system. In a tungsten inert gas (TIG)-based process, the torch electrode wears out over time and must be changed to avoid defective deposition. This paper proposes an approach for a compact torch end-effector to improve the robot’s build volume. This paper also proposes an approach to reducing non-printing process time by designing and implementing a semi-automated electrode changing system.

Yaw error is the angle between a turbine’s rotor central axis and the wind flow. The presence of yaw error results in lower power production from turbines. Yaw error also puts extra loads on turbine components, which in turn, lowers their reliability. In this study we develop a stochastic model to calculate the average capacity factor of a 50 turbine offshore wind farm and investigate the effects of minimizing the yaw error on the capacity factor. In this paper, we define the capacity factor in terms of energy production, which is consistent with the common practice of wind farms (rather than the power production capacity factor definition that is used in textbooks and research articles). The benefit of using the energy production is that it incorporates both the power production improvements and downtime decreases. For minimizing the yaw error, a nacelle mounted LIDAR is used. While the LIDAR is on a turbine, it collects wind speed and direction data for a period of time, which is used to calculate a correction bias for the yaw controller of the turbine, then it will be moved to another turbine in the farm to perform the same task. The results of our investigation shows that although the improvements of the capacity factor are less than the theoretical values, the extra income from the efficiency improvements is larger than the cost of the LIDAR.

Boiling influences many industrial processes like quenching, desalination and steam generation. Boiling heat transfer at high temperatures is limited by the formation of a vapor layer between the solid and fluid. Low thermal conductivity of this vapor layer inhibits heat transfer. Electrowetting (EW) fields can breakdown this vapor layer to promote wetting, and this concept works for many quenching media including water and organic solvents. This work studies the suppression of this vapor layer and measures the resulting heat transfer enhancement during quenching of metals. We image the fluid-surface interactions and boiling patterns in the presence of an electrical voltage. EW fields replace film boiling with periodic wetting-rewetting cycles and thus fundamentally change the heat transfer mode. The increased wettability substantially reduces the cool down time. The cooling rate can by increased by as much as 3X. The results show that electric fields can dynamically tune the classical quenching curve. This study opens up new avenues to control the metallurgy of metals via electrical control of the cooling rate.

Axial piston variable displacement pumps (VDAP) are the heart of every hydraulic system, and are they commonly used in the industry for its high capacity, efficiency (volume and total), and good performance in the handling of high pressures and speeds. Faults are usually associated with wear and leakage processes, which cause significant decreases in performance. This paper discusses about the advantages to implement a Condition-based Maintenance (CBM) and the use of techniques of fault diagnosis in axial variable displacement piston pumps (VDAP), as they are: Neural Networks (NN), Support Vector Machine (SVM) and Fuzzy Logic (FL) and other hybrid models. The results of this investigation provide guidelines for the selection of the most suitable technique to prevent faults in VDAP in order to help reducing down time, increase productivity and competitiveness of companies that requires the use of fluid power systems by enabling the implementation of a non-intrusive fault diagnosis management system, which must be reliable, economical and easily accessible to the industry.

The tasks of maintenance and repair without optimal planning can be costly and result in prolonged maintenance times, reduced availability and possible flight delays. Aircraft manufacturers and maintainers see significant benefits in constantly improving Health Management and Maintenance (HMM) practices by deploying the most effective maintenance planning strategies. The planning of the maintenance and repair is a complex task due to chain dependency of engines to aircraft, and aircraft to the flight schedules. This paper presents a scheduling method for determining the time of maintenance based on the historical engine operation data in order to maximize the use of estimated remaining useful life of the engines as well as lowering the cost and duration of the downtime. The Time-on-Wing (TOW) data is used in conjunction with probability density functions to determine the shape of the respective distribution of the time of maintenance to minimize the loss of expected remaining useful life. Data from each engine with most chance of failure is then selected and fed into an extended Branch and Bound (B&B) routine to determine the best optimum sequence for entering the facility in order to minimize the waiting time.

Gas compressor reliability is vital in oil and gas industry because of the equipment criticality which requires continuously operations. Currently, plant operators often face difficulties in predicting appropriate time for maintenance and would usually rely on time-based predictive maintenance intervals as recommended by Original Equipment Manufacturer. Delayed decision on compressor maintenance intervention has caused prolonged downtime due to poor readiness of spare parts and resources. The paper discussed on development of a software-based tool which is able to assist machinery engineers to quantify performance deterioration of gas compressor and predict optimum time for maintenance activities. Maintenance history data is collected and analysed regularly and maintenance advices are subsequently produced based on the input parameters. Natural gas compressors of an oil producing offshore platform at Peninsular Malaysia are used as a case study of this project. It was found that isentropic efficiency and head decrease, but gas power increases parabolically with time for the low pressure compressors, suspected due to heavy component fouling. From these information, Compressor Performance Monitoring Program is developed which able to compute the fouling level of the compressor in terms of performance indicators deviations. The results are then being utilized to estimate future maintenance requirements based on historical data. In general, this software provides a powerful tool for gas compressor operators to realize predictive maintenance approach in their operations.

Risk assessment is used in many industries to identify and manage risks. Initially developed for use on aeronautical and nuclear systems, risk assessment has been applied to transportation, chemical, computer, financial, and security systems among others. It is used to gain an understanding of the weaknesses or vulnerabilities in a system so modification can be made to increase operability, efficiency, and safety and to reduce failure and down-time. Risk assessment results are primary inputs to risk-informed decision making; where risk information including uncertainty is used along with other pertinent information to assist management in the decision-making process. Therefore, to be useful, a risk assessment must be directed at specific objectives. As the world embraces the globalization of trade and manufacturing, understanding the associated risk become important to decision making. Applying risk assessment techniques to a global system of development, manufacturing, and transportation can provide insight into how the system can fail, the likelihood of system failure and the consequences of system failure. The risk assessment can identify those elements that contribute most to risk and identify measures to prevent and mitigate failures, disruptions, and damaging outcomes. In addition, risk associated with public and environment impact can be identified. The risk insights gained can be applied to making decisions concerning suitable development and manufacturing locations, supply chains, and transportation strategies. While risk assessment has been mostly applied to mechanical and electrical systems, the concepts and techniques can be applied across other systems and activities. This paper provides a basic overview of the development of a risk assessment.

Gearboxes have been prone to early failure rather than any mechanical part of modern wind turbines, much earlier than their predicted design life. Some studies indicated that gearboxes of wind turbines fail during the first 3 to 5 years of operation of the system as opposed to the total design life of the wind turbine, which usually is 20 years. Consequently, such failures cause the highest down time and extremely expensive replacement activities. Gearboxes are subjected to torsional, bending and axial wind loads which are yet not fully defined. The uncertainty in loading conditions and system design parameters has brought about the importance of considering probabilistic design and modeling approach than the traditional deterministic approach. Accordingly, the motivation of this study is to improve the reliability of gearboxes for wind turbine applications. A probabilistic multibody dynamic modeling of the gearbox, that fully integrates uncertainties in wind loading and design parameters, is sought. This paper presents previous studies and finally proposes the above mentioned approach as a potential way of improving, in general, the reliability of wind energy and, in particular, the gearboxes in wind turbines.

Induction motors are the workhorses of industry and a lot of effort has been invested in detecting and diagnosing induction motor faults through the analysis of the motor electrical signals. However, in many industrial applications, electric motors are used to drive dynamic loads such as pumps, fans, blowers etc. Failure of either the motors or the driven loads is associated with operational disruption. Consequently it would be beneficial if the entire motor-pump system is monitored and diagnosed. The large costs associated with production losses can be avoided if system degradation can be detected at early stages prior to failure. Moreover, downtime can be further reduced if the faulty component within the drive power system can be isolated thereby aiding plant personnel to be better prepared with spares and repair kits. Hence there is not only a strong need for cost-effective detection schemes to assess the condition of the drive power system as a whole, but also a strong need for efficient isolation schemes to identify the component within the system that is faulty. This paper describes a sensorless approach to detect and isolate induction motor and/or centrifugal pump faults. Motor and pump bearing degradation is considered to validate the performance effectiveness of the proposed scheme. No add-on sensors, on either the motor or the pump, are used in the development of the proposed method to avoid any reduction in overall system reliability and prevent increased costs. In fact, motor and/or pump bearing degradation is detected and isolated using only the motor line voltages and phase currents. The proposed technique is insensitive to electric power supply fluctuations and mechanical load variations and it does not require prior knowledge of either the motor or the pump design parameters. Hence this approach can be easily ported to motor-pump systems of varying manufacturers and sizes. The developed algorithm has been tested on accelerated fault data collected from a centrifugal pump fluid loop driven by a 3-φ, 3 hp induction motor. Results from these experiments indicate that the proposed fault detection and isolation scheme successfully detects and classifies bearing degradation in the motor and/or the pump without false positives or misclassification.

Electrical signal analysis has been in use for quite some time to detect and diagnose induction motor faults. In most industrial applications, induction motors are used to drive dynamic loads such as pumps, compressors, fans, etc. Failure of either the motors or the driven loads results in an unscheduled downtime which in turn leads to loss of production. These operational disruptions could be avoided if the equipment degradation is detected in its early stages prior to reaching catastrophic failure conditions. Hence the need arises for cost-effective detection schemes not only for assessing the condition of electric motors but also the driven loads. This paper presents an experimentally demonstrated sensor-less approach to detect impeller cracks in centrifugal pumps. The proposed method is sensorless in the sense that it does not use any mechanical and/or process pump sensors to detect impeller faults. Rather motor electrical measurements are used for the intended purpose. Mechanical sensors have high costs and low reliability, and frequently fail more often than the system being monitored. Hence add-on mechanical sensors reduce the overall system reliability. In this study, fault detection is accomplished using only the line voltages and phase currents of the electric motor driving the pump. The developed detection algorithm is insensitive to electrical power supply and load variations. Furthermore, it does not require prior knowledge of either a motor or the pump model or design parameters and hence the detection algorithm can be easily ported to motor-pump systems of varying manufacturers and sizes. The proposed fault detection scheme has been tested on data collected from a centrifugal pump driven by a 3-φ, 3 hp induction motor. Several cracks on the pump impeller are staged to validate the detection effectiveness of the proposed scheme and compare its effectiveness with respect to continuous vibration monitoring. In addition to these staged faults, experiments are also conducted to demonstrate the prevention of false alarms by the algorithm. Results from all of these experiments are presented to substantiate the performance of the sensorless pump fault detection scheme.

With the increased complexity of today’s industrial processes, maintaining equipment by preventing unscheduled downtime using monitoring hardware is a key challenge. Industrial statistics indicate that seal and impeller failures are predominant failure modes in centrifugal pumps and they are not adequately addressed in the literature. In this paper, a neural network (NN) based Nonlinear Autoregressive Moving Average with Exogenous input (NARMAX) model is used to develop fault detection scheme for detecting seal and impeller failures in centrifugal pumps. A rigorous methodology of detecting failures at the incipient stage is introduced. First a nonlinear relationship among the monitored parameters (inlet and outlet pressure, outlet flow, inlet and outlet temperature, and acceleration) where the previous values of the indicative parameters are used as inputs to the NARMAX model and the output being the value at the current instance is captured. The NARMAX modeled outputs are compared with the actual measured values in order to generate residuals. By choosing a suitable threshold, we could minimize false and missed alarms. Mathematical procedure for selection of threshold is derived in this paper. Along with the NARMAX model, an online approximator is used in the fault detection scheme for understanding the faults in the system. Experiments on the centrifugal pump seal and impeller failures were conducted by using a laboratory test bed. Experimental results show that the proposed fault detection scheme is able to successfully detect failures.

This paper provides an introduction to various safety requirements for the design, manufacture and use of industrial machinery systems. To be successful, manufacturers must be aware of the different safety requirements that exist in North America and Europe with respect to electrical, mechanical and electromagnetic compatibility. With this knowledge, manufacturers can reduce the overall time to market and curb potential lawsuits. Users will also benefit from the safer equipment with a reduction of work-related accidents, injuries and down-time.

In today’s semiconductor industry, the business loss associated with one full day of lost production can have a significant impact on revenues. Key factors affecting overall profitability are unplanned fabrication downtimes and contamination resulting in defective products. It is critical, therefore, to implement effective risk management strategies for the fabrication process to eliminate yield-limiting defects and to improve the overall production. This paper will discuss a study completed for a semiconductor chemical supplier. Battelle applied a methodology based on the Process Event Tree/Hazard and Operability Study (PET-HAZOP) methodology, identifying numerous ways to avoid unnecessary outages and lost production at a specific client’s semiconductor fabrication facility. This methodology uses a systematic approach to identifying risks in a process, and risks are quantified in terms of expected dollars per year. Within the analysis, the importance of individual equipment items in the process is measured with respect to risk. Two risk measures are used in the analysis: Risk Reduction Worth (RRW) and Risk Achievement Worth (RAW). The RRW identifies opportunities for areas of risk reduction and the RAW identifies critical equipment that must be maintained to prevent the overall risk of the system from increasing significantly. Results were provided in a cost-benefit form for the top items on the RRW and RAW lists. For example, replacing existing chemical distribution pumps with completely non-metallic pumps (cost of about $60,000) would result in savings, due to avoidance of possible chemical contamination, of $430,000 per year. This, and other findings, proved that a systematic approach to evaluating risks for the semiconductor front-end process can identify valuable risk reduction alternatives resulting in significant cost savings.

The selection of appropriate monitoring processes is an important decision to be made. Monitoring of manufacturing processes plays a very important role to avoid down time of the machine, or prevent unwanted conditions such as chatter, excessive tool wear or breakage. Most monitoring systems developed up to date employ force, acoustic emission and vibration, or a combination of these and other techniques with a sensor integration strategy. In this work, the implementation of a monitoring system utilizing simultaneous vibration and strain measurements on the tool tip is investigated for the average flank wear of coated carbide tools which are used in finishing turning process, with cast iron shaft as a work-piece. Data from the manufacturing processes were recorded with one piezoelectric strain sensor and an accelerometer, each coupled to the data acquisition card. There are 24 features indicative of tool wear were extracted from the original signal. These include features from the time domain, frequency domains, time-series model coefficients and four packet features extracted from wavelet packet analysis. The (2 × 3) self organizing map (SOM) neural network was employed to identify the tool state as a result of the present work, the tool wear classified by applied two independent wear test as a training tests, and checked the results of the SOM by applied an independent test, finally we have an SOM model can classifying the tool wear with minimal error.

The use of diagnostics as a maintenance technology for motion systems can play a key role for the life cycle costs of a production facility. A new comprehensive diagnostic approach provides methods for pneumatic systems that continuously monitor and analyze sensor readings in the main supply line and within subsystems. Using different evaluation methods of pattern recognition, the source of deviations can be isolated to individual components. Unlike traditional component-based diagnostic solutions, this new approach is applicable to complete pneumatic systems and/or single pneumatic components for sensing and monitoring performance levels, information acquisition for optimized maintenance preventing unscheduled downtime and extending machine or process serviceability. In a comprehensive cost calculation, the cost savings over the life cycle of a production facility of an end user are revealed.