Search

Fault detection and diagnosis (FDD) in refrigeration systems is of great importance for ensuring better equipment reliability and energy efficiency. Although numerous researches studied FDD algorit...

Effective chiller fault diagnosis is of great importance for maintaining a better service and energy efficiency. Deep learning proficiently solves some problems challenging Artificial Intelligence and becomes one of the excellent candidates for fault diagnosis recently. This study proposes a novel fault diagnosis strategy for a chiller, which merges simulated annealing (SA) into a deep neural network (DNN) to obtain effective, efficient, and stable performance. The proposed SA-DNN strategy is carefully compared with DNN and back-propagation (BP) network. The results show that SA-DNN enhances the diagnostic accuracy, shortens the running time, and greatly improves the model stability. The optimal network structure has 2 hidden layers (HL) with each layer 64 nodes, and the overall diagnostic accuracy for seven typical faults attains 99.30%. The nodes in the first HL are proved to be dominant over those in the second or behind because the mapping of the second can hardly make corrections if that of the first is deformed already. The global faults are hard to be diagnosed due to the global effect, but the proposed strategy achieves satisfactory results with the highest individual accuracy reaching 99.79% for excess oil and the lowest 97.52% for refrigerant leakage. The features used for diagnosis have an influence on the accuracy of the proposed method.

The existing fault detection and diagnosis (FDD) model of chillers requires considerable normal and fault data. The acquisition of these data is time-consuming and expensive, and the model is only ...

Graphene nanosheet have an ultra-thin sheet structure that can easily fill the contact friction surfaces and reduce friction losses. Its thermal conductivity reaches 5,300 W/mK. Therefore, a nanofluid prepared by adding graphene nanosheets to base liquid will have lower coefficient of friction and higher thermal conductivity. In this study, 1:1 mass ratio of graphene nanosheets additive to Span85 dispersant was added to SUNISO 3GS refrigerant oil, using ultrasonic oscillation dispersing and chemical methods to prepare 10 mg/L, 20 mg/L and 30 mg/L graphene nanosheets refrigerant oil. The cooling rate, freezing capacity, and energy consumption of the refrigerator using different concentrations of graphene nanosheets refrigeration oil were compared. The performance of the compressor and the refrigerator were analyzed. The results showed that during the stable refrigerator operation, adding graphene nanosheets to lubricants for the refrigerator had lower suction temperature, discharge temperature, condenser temperature, suction pressure and discharge pressure than pure oil. Both the fresh food compartment and freezer's cooling rates were increased. The increase of the concentration leaded to better freezing capacity of the freezer. The 30 mg/L graphene nanosheets refrigerant oil increased the freezing capacity of the refrigeration system by 11.1%. The on-time ratio, operating current of the compressor and energy consumption were also reduced. Compared with pure oil, the energy consumption in 24 h of the domestic refrigerator using 10 mg/L, 20 mg/L and 30 mg/L graphene nanosheets refrigerant oil dropped by 15.4%, 19.2% and 20.3%.

The heat transfer performance of oscillating heat pipes (OHPs) with graphene nanoplatelet (GNP) nanofluids was investigated experimentally. In these experiments, the GNP nanofluid concentrations were 1.2, 2.0, 5.7, 9.1, 13.8, and 16.7 vol%, the heating power ranged from 10 to 100 W, and the filling ratios were 45%, 55%, 62%, 70%, and 90%. The results indicate that the heat transfer performance of OHPs is improved by using GNP nanofluids as the working fluid compared to an OHP with deionized water (DW). At appropriate filling ratios (55%, 62%, and 70%), the optimum range of GNP nanofluid concentrations was 2.0–13.8 vol%. The maximum reduction in thermal resistance was 83.6% for an OHP filled with a 2.0 vol% GNP nanofluid compared to the DW OHP at the same filling ratio (62%) and heating power (80 W). At a low filling ratio (45%), adding GNP nanofluids to DW could alleviate dry-out and improve the heat transfer performance of the OHP.

Segment routing is an emerging routing technology that was initially driven by commercial vendors to achieve scalable, flexible, and controllable routing. In segment routing, multiple multi-protocol label switch labels are stacked in the packet header to complete end-to-end transmission, which may lead to a large label stack and a long packet header. Thus, scalability issues may occur when segment routing is applied to large-scale networks. To address this issue, multiple mechanisms and algorithms have been proposed for minimizing the label stack size. However, we argue that these methods ignore the constraint on the maximum segment list depth (SLD), since the typical network equipment can currently only support three to five layers of labels. In this paper, we study segment routing with the maximum SLD constraint and demonstrate that issues, such as explosive increases in the size of the label space and the management overheads will arise when the maximum SLD constraint is imposed. To address these issues, we make contributions from two main aspects. First, based on the network programmability that is provided by openflow, a novel segment routing architecture with improved data plane is proposed that reduces the overhead of additional flow entries and label space. Second, a new path encoding scheme is designed to minimize the SLD under the given maximum constraint, while taking multiple types of overhead into consideration. Moreover, we also perform simulations under different scenarios to evaluate the performances of the proposed algorithms. The simulation results demonstrate that the proposed mechanisms and algorithms can address the issues of segment routing when there is a constraint on the maximum SLD.

In this paper, an experimental study is presented on the thermal resistance characteristics of closed loop pulsating heat pipes (CLPHPs) with methanol-based binary mixtures. The working fluids were methanol mixed with deionized water, acetone and ethanol. The volume mixing ratios used were 2:1, 4:1 and 7:1, and the heating power ranged from 10 W to 100 W with filling ratios of 45%, 62%, 70% and 90%. The results showed that adding other working fluids to methanol could change the thermal resistance characteristics of a PHP. At a low filling ratio (45%), adding water to methanol could prevent dry-out at a high heating power; when ethanol was added to methanol, the thermal resistance of the CLPHP was between that with pure methanol and ethanol; when acetone was added, the thermal resistance of the CLPHP was slightly lower than that with pure methanol and acetone. At a high filling ratio (62%, 70%, 90%), the thermal resistance characteristics of CLPHPs with methanol based mixtures were not much different from those with pure fluids except for methanol–water mixture where the thermal resistance was greater than that with pure methanol and pure water. It can be inferred that the heat transfer performances of CLPHPs with methanol-based binary mixtures are related to the thermal-physical properties of the working fluids, vapor–liquid phase transition properties, molecular interactions and the additional resistance to mass transfer.

The heat transfer performances of a pulsating heat pipe (PHP) with ethanol–water, ethanol–methanol and ethanol–acetone are investigated experimentally. The mixing ratios (MRs) of the ethanol-based mixed working fluids are 2:1 and 4:1, the volume filling ratios (FRs) range from 45% to 90% and the heat input ranges from 10 W to 100 W. The experimental results are as follows: When the mixing ratio is 2:1, the heat transfer performance of PHP with ethanol–water is better than other working fluids at a filling ratio of 45% because of the phase-change inhibition in ethanol–water; at a filling ratio of 55%, PHP with ethanol–acetone shows better performance among those with mixed working fluids. Acetone with a relatively high value of ( dp / dT ) sat (saturation pressure gradient versus temperature) and relatively lower dynamic viscosity can lead to relatively high velocity in the PHP, which can decrease the temperature difference between the evaporation section and condensation section. When the mixing ratio is 4:1, the thermal resistance of PHP with ethanol–water that is close to being dried out under a filling ratio of 45% rises faster than that at a mixing ratio of 2:1 due to the presence of less deionized water (DI water); the heat transfer performance of PHP with ethanol–acetone is excellent at a filling ratio of 55% among the ethanol-based mixed working fluids because the thermal resistance is relatively small and the maximum heat input that PHP can endure is highest. When the volume filling ratio reaches 62%, 70% and 90%, the heat transfer performance of PHP with pure working fluids is better than that with ethanol-based mixed working fluids. This may be partially attributed to the offset of flow driving force caused by the mass transfer due to the concentration difference between the liquid and the vapour phase of mixtures. The filling ratio of 62% shows a marginal leading in terms of lower thermal resistance.

Considering the small cooling capacity of a Hampson-type Joule-Thomson (J-T) cooler, a multi-layer printed circuit board (PCB) J-T cooler with parallel microchannels was fabricated and investigated in this study. The cooler has four parts: an inlet, a counter flow heat exchanger, a throttle, and an evaporator. A mathematical model is established and iteratively solved with a developed code. Using Ar as the working fluid, the cooling characteristics of the J-T cooler at 2–8 MPa were simulated. The pressure, temperature and velocity of the fluids as well as the temperature of the materials were obtained along the length of the cooler. The discrepancy between calculation and experiment of the cold end temperature is within ±2%, and the discrepancy of the outlet pressure is within ±4%, indicating the reasonability and reliability of the mathematical model. The model was also used to obtain the heat transfer rate between high- and low-pressure fluids, the enthalpy change, the cooling capacity of the cooler and the axial heat conduction of the material. The results show that the cold end temperature can reach 155.9 K with the gross cooling capacity being 3.66 W at the inlet pressure of 8.02 MPa. It was found that if the axial heat conduction is omitted, the cold end temperature could decrease further by 2.2 K. Finally, the optimisation measures of the cooler are discussed and proposed by analysing the variation law of the J-T coefficient.

For better service and energy savings, improved fault detection and diagnosis (FDD) of building energy systems is of great importance. To achieve this aim, ensemble learning is investigated and introduced in this study. Three types of methods like K-nearest neighbor (KNN), support vector machine (SVM), and random forest (RF) are carefully selected, optimized, and integrated into an ensemble diagnostic model (EDM) by the majority voting method. Experimental data for seven typical gradual faults in a centrifugal building chiller are used for model validation and evaluation. The results show that the diagnostic accuracy of the EDM (99.88%) is higher than that of the individual methods, with significant improvements for normal operation and refrigerant leakage, and no false alarms reported. Models based on ensemble learning, EDM and RF (homogenous ensemble), exhibit better performance for global faults, which are difficult to diagnose. In addition, five different feature sets are selected from the literature for further tests. It is found that the diagnostic performance depends not only on the principle of diagnosis, but also on the fault category and the characteristics of the feature set such as the indicative degree to corresponding faults, number of features, correlation degree between features, and information redundancy, and feature selection is proved to be more important than algorithm selection in fault diagnosis practice. Ensemble learning is proved to be a promising candidate for the fault diagnosis of building energy systems, except for the Zhou-8 feature set (eight temperature features), for which KNN is a better choice.

Despite the importance of fault diagnosis in refrigeration systems, the performance and improvement of most existing diagnostic models are limited by their reliance on a single method. This study proposes a novel application of ensemble learning that incorporates several intelligent ensemble members into an integrated model by means of majority voting. The ensemble members include k-nearest neighbour (KNN), support vector machine (SVM), decision tree (DT), random forest (RF) and logistic regression (LR). ASHRAE fault data were employed to establish the model. In addition, this study explores the integration of different subsets of ensemble members and revealed that the optimum subset combination comprised KNN, DT, and RF. Although the accuracy of this model was slightly lower than the model with all five ensemble members, it realised a substantial reduction in runtime. Compared to an SVM optimization model, the integrated model realised higher accuracy and reduced training time, without requiring parameter optimization. This achievement merits note as ensemble learning is traditionally associated high time-costs. Further investigation revealed that both the diversity and high accuracy of ensemble members are required to obtain an effective integrated model. These observations demonstrate the proposed model offers a promising alternative solution for fault diagnosis in refrigeration systems.

As noted in this investigation, there are only eight sensors commonly installed in factory chillers. However, the fault detection and diagnosis (FDD) models of chillers are now mostly based on laboratory data. In the laboratory, the unit is equipped with more sensors than commercial machines are. The FDD results obtained by this method does not match the data from the factory chillers and cannot be applied in practice. Based on the ASHRAE RP-1043 data, this paper extracts information from three of the factory-installed (FI) sensors, along with information from all eight of the FI sensors, in order to establish the support vector machine (SVM)-3 and SVM-8 diagnostic models based on grid search and cross validation parameter optimization for the seven typical faults. These models are compared to the SVM-64 model, which is based on the 64 parameters from the ASHRAE data. The overall accuracy of SVM-8 model is 97.68%, which meets the needs of field operation diagnosis. In addition, based on the eight factory-installed sensors, if the pressure of oil feed and temperature of oil feed sensors are added together to diagnose the fault, the normal detection rate and the diagnostic performance for condenser fouling and excess oil faults improve, and the overall fault diagnosis is more effective.

Data-driven diagnostic models for refrigeration systems are often used exclusively for a dedicated object. For other types of chillers, a new model must be trained using normal and faulty data, which is both time consuming and heavily data dependent, and accordingly, curbs its application. In this study, the technology for handling imbalanced data is introduced and combined with support vector machine (SVM) to probe the possibility of transferring the FDD (fault detection and diagnosis) knowledge of a centrifugal chiller to a screw chiller by using just a small amount of new data. Principal component analysis (PCA) and the synthetic minority oversampling technique (SMOTE) were used to oversample the faulty sample set with an imbalance ratio of 5%, and a support vector machine (SVM) was employed for fault diagnosis. The experimental results indicate that by using PCA-SMOET-SVM technology, the overall diagnostic performance of the screw chiller with much less data/information is improved with the aid of the prior knowledge transferred from the centrifugal chiller. By investigating the oversampling ratios between 100% and 400%, it was found that the ratio of 100% was the best with the average diagnostic accuracy reaching 96.70% for the faults of the screw chiller.

With the increasing miniaturization of high-powered electronic components, technologies for rapid cooling in small spaces have received more and more attention. Joule-Thomson (J-T) refrigeration is an effective means for rapid cooling. It is widely used in military and medical refrigeration, electronic equipment cooling and many other fields. This paper describes the structural development of J-T refrigerators of the traditional Hampson type, etched microchannel type and others. It reviews experimental studies on refrigeration components and the charge composition of the mixtures in the J-T cryogenic system. It also introduces the main applications of the J-T cryogenic system. Finally, it summarizes the above research and puts forward developmental prospects, hoping to provide reference for future experimental design and research.

Actuation system is a vital system in an aircraft, providing the force necessary to move flight control surfaces. The system has a significant influence on the overall aircraft performance and its safety. In order to further increase already high reliability and safety, Airbus has implemented a dissimilar redundancy actuation system (DRAS) in its aircraft. The DRAS consists of a hydraulic actuation system (HAS) and an electro-hydrostatic actuation system (EHAS), in which the HAS utilizes a hydraulic source (HS) to move the control surface and the EHAS utilizes an electrical supply (ES) to provide the motion force. This paper focuses on the performance degradation processes and fault monitoring strategies of the DRAS, establishes its reliability model based on the generalized stochastic Petri nets (GSPN), and carries out a reliability assessment considering the fault monitoring coverage rate and the false alarm rate. The results indicate that the proposed reliability model of the DRAS, considering the fault monitoring, can express its fault logical relation and redundancy degradation process and identify potential safety hazards.