Spark Configuration Research Articles

The effects of water injection inside an internal combustion engine regarding NOx emissions

Abstract In this paper we present the influence of water injection on NOx emissions and the performance of a gasoline engine subjected under light to medium loads. The test bench used for measurements has a 4-cylinder, 1.6-liter engine and operates with a wasted spark configuration. Water is injected upstream of the throttle valve through a single injector mounted in the intake runner. The injection timing and amount of water injected inside the engine have been controlled by NI CompactRIO hardware device by the aid of an in-house algorithm implemented in LabView. Water is injected twice at each rotation of the crankshaft in order to have water in all cylinders. The injection of water is done at a fix angle, one time at 0°CA and the other at 360°CA, only the amount is varied from 10% to 45% of the total fuel fraction injected. The results showed that at an amount of water of about 25% of the total fraction of fuel injected, it produced a considerable decrease in NOx emissions, while keeping engine performance unchanged.

PSR J0026−1955: A curious case of evolutionary subpulse drifting and nulling

ABSTRACT PSR J0026−1955 was independently discovered by the Murchison Widefield Array recently. The pulsar exhibits subpulse drifting, where the radio emission from a pulsar appears to drift in spin phase within the main pulse profile, and nulling, where the emission ceases briefly. The pulsar showcases a curious case of drift rate evolution as it exhibits rapid changes between the drift modes and a gradual evolution in the drift rate within a mode. Here, we report new analysis and results from observations of J0026−1955 made with the upgraded Giant Metrewave Radio Telescope at 300–500 MHz. We identify two distinct subpulse drifting modes: A and B, with mode A subcategorized into A0, A1, and A2, depending upon the drift rate evolutionary behaviour. Additionally, the pulsar exhibits short and long nulls, with an estimated overall nulling fraction of ∼58 per cent, which is lower than the previously reported value. Our results also provide evidence of subpulse memory across nulls and a consistent behaviour where mode A2 is often followed by a null. We investigate the drift rate modulations of J0026−1955 and put forward two different models to explain the observed drifting behaviour. We suggest that either a change in polar gap screening or a slow relaxation in the spark configuration could possibly drive the evolution in drift rates. J0026−1955 belongs to a rare subset of pulsars which exhibit subpulse drifting, nulling, mode changing, and drift rate evolution. It is therefore an ideal test bed for carousel models and to uncover the intricacies of pulsar emission physics.

Hybrid SWOC-AHP Analysis of Two-Stroke Twin Spark Engine

A two-stroke engine is a promising device that provides a high power-to-weight ratio in automobiles. In recent times, the twin-spark configuration has shown advantages over the single-spark configuration. However, the rapid combustion associated with the twin-spark configuration leads to a higher NO x emission. Therefore, it is important to determine its strength-weakness-opportunity-challenges (SWOC). It is also important to prioritize the SWOC factors and sub-factors connected with the twin spark two-stroke engine. Thus, the primary aim of the current investigation is to perform the hybrid SWOC-analytical hierarchy process (AHP) analysis for producing the quantifiable priorities for the factors of the two-stroke twin spark configuration. Additionally, the rank-stability of SWOC factors is investigated using sensitivity analysis. From the results, challenge factors emerged to be the most significant, and to counter the challenges, the deployment of opportunity factors showed an improvement in the performance of the twin spark configuration. Moreover, weaknesses attained the least position, which shows a positive perception for developing the future emissions-compliant engine designs. Finally, implementing a twin spark configuration in a two-stroke engine can play a significant role in enhancing its additional benefits.

Phronesis: Efficient Performance Modeling for High-dimensional Configuration Tuning

We present Phronesis, a learning framework for efficiently modeling the performance of data analytic workloads as a function of their high-dimensional software configuration parameters. Accurate performance models are useful for efficiently optimizing data analytic performance. Phronesis explicitly considers the error decomposition in statistical learning and implications for efficient data acquisition and model growth strategies in performance modeling. We demonstrate Phronesis with three popular machine learning models commonly used in performance tuning: neural network, random forest, and regression spline. We implement and evaluate it for Spark configuration parameters. We show that Phronesis significantly reduces data collection time for training predictive models by up to 57% and 37%, on average, compared to state-of-the-art techniques in building Spark performance models. Furthermore, we construct a configuration autotuning pipeline based on Phronesis. Our results indicate up to 30% gains in performance for Spark workloads over previous, state-of-the-art tuning strategies that use high-dimensional models.

A Novel Reinforcement Learning Approach for Spark Configuration Parameter Optimization.

Apache Spark is a popular open-source distributed data processing framework that can efficiently process massive amounts of data. It provides more than 180 configuration parameters for users to manually select the appropriate parameter values according to their own experience. However, due to the large number of parameters and the inherent correlation between them, manual tuning is very tedious. To solve the problem of tuning through personal experience, we designed and implemented a reinforcement-learning-based Spark configuration parameter optimizer. First, we trained a Spark application performance prediction model with deep neural networks, and verified the accuracy and effectiveness of the model from multiple perspectives. Second, in order to improve the search efficiency of better configuration parameters, we improved the Q-learning algorithm, and automatically set start and end states in each iteration of training, which effectively improves the agent’s poor performance in exploring better configuration parameters. Lastly, comparing our proposed configuration with the default configuration as the baseline, experimental results show that the optimized configuration gained an average performance improvement of 47%, 43%, 31%, and 45% for four different types of Spark applications, which indicates that our Spark configuration parameter optimizer could efficiently find the better configuration parameters and improve the performance of various Spark applications.

Spark Configurations to Optimize Decision Tree Classification on UNSW-NB15

This paper looks at the impact of changing Spark’s configuration parameters on machine learning algorithms using a large dataset—the UNSW-NB15 dataset. The environmental conditions that will optimize the classification process are studied. To build smart intrusion detection systems, a deep understanding of the environmental parameters is necessary. Specifically, the focus is on the following environmental parameters: the executor memory, number of executors, number of cores per executor, execution time, as well as the impact on statistical measures. Hence, the objective was to optimize resource usage and minimize processing time for Decision Tree classification, using Spark. This shows whether additional resources will increase performance, lower processing time, and optimize computing resources. The UNSW-NB15 dataset, being a large dataset, provides enough data and complexity to see the changes in computing resource configurations in Spark. Principal Component Analysis was used for preprocessing the dataset. Results indicated that a lack of executors and cores result in wasted resources and long processing time. Excessive resource allocation did not improve processing time. Environmental tuning has a noticeable impact.

On the scalability of Big Data Cyber Security Analytics systems

Big Data Cyber Security Analytics (BDCA) systems use big data technologies (e.g., Apache Spark) to collect, store, and analyse a large volume of security event data for detecting cyber-attacks. The volume of digital data in general and security event data in specific is increasing exponentially. The velocity with which security event data is generated and fed into a BDCA system is unpredictable. Therefore, a BDCA system should be highly scalable to deal with the unpredictable increase/decrease in the velocity of security event data. However, there has been little effort to investigate the scalability of BDCA systems to identify and exploit the sources of scalability improvement. In this paper, we first investigate the scalability of a Spark-based BDCA system with default Spark settings. We then identify Spark configuration parameters (e.g., execution memory) that can significantly impact the scalability of a BDCA system. Based on the identified parameters, we finally propose a parameter-driven adaptation approach, SCALER, for optimizing a system's scalability. We have conducted a set of experiments by implementing a Spark-based BDCA system on a large-scale OpenStack cluster. We ran our experiments with four security datasets. We have found that (i) a BDCA system with default settings of Spark configuration parameters deviates from ideal scalability by 59.5% (ii) 9 out of 11 studied Spark configuration parameters significantly impact scalability and (iii) SCALER improves the BDCA system's scalability by 20.8% compared to the scalability with default Spark parameter setting. The findings of our study highlight the importance of exploring the parameter space of the underlying big data framework (e.g., Apache Spark) for scalable cyber security analytics.

Revisiting the subpulse drifting phenomenon in PSR J1822−2256: drift modes, sparks, and emission heights

ABSTRACT Subpulse drifting in pulsar radio emission is considered to be one of the most promising phenomena for uncovering the underlying physical processes. Here, we present a detailed study of such a phenomenon in observations of PSR J1822−2256, made using the upgraded Giant Meterwave Radio Telescope (uGMRT). Observations were made simultaneously using the band 3 (300–500 MHz) and band 4 (550–750 MHz) receivers of the uGMRT. The pulsar is known to exhibit subpulse drifting, mode changing, and nulling. Our observations reveal four distinct subpulse drifting modes of emission (A, B, C, and D) for this pulsar, with the drift periodicities of 17.9P1, 5.8P1, 8P1, and 14.1P1, respectively (where P1 is the pulsar rotation period), two of which exhibit some new features that were not reported in the previous studies. We also investigate the possible spark configuration, characterized by the number of sparks (n) in the carousel patterns of these four drift modes, and our analysis suggests two representative solutions for the number of sparks for a carousel rotation period, P4, which lies in the range of 13–16. The large frequency coverage of our data (300–750 MHz) is also leveraged to explore the frequency dependence of single-pulse characteristics of the pulsar emission, particularly the frequency-dependent subpulse behaviour and the emission heights for the observed drift modes. Our analysis suggests a clear modal dependence of inferred emission heights. We discuss the implications for the pulsar emission mechanism and its relation to the proposed spark configuration.

You Only Run Once: Spark Auto-Tuning From a Single Run

Tuning configurations of Spark jobs is not a trivial task. State-of-the-art auto-tuning systems are based on iteratively running workloads with different configurations. During the optimization process, the relevant features are explored to find good solutions. Many optimizers enhance the time-to-solution using black-box optimization algorithms that do not take into account any information from the Spark workloads. In this article, we present a new method for tuning configurations that uses information from one run of a Spark workload. To achieve good performance, we mine the SparkEventLog that is generated by the Spark engine. This log file contains a large amount of information from the executed application. We use this information to enhance a performance model with low-level features from the workload to be optimized. These features include Spark Actions, Transformations, and Task metrics. This process allows us to obtain application-specific workload information. With this information our system can predict sensible Spark configurations for unseen jobs, given that it has been trained with reasonable coverage of Spark applications. Experiments show that the presented system correctly produces good configurations, while achieving up to 80% speedup with respect to the default Spark configuration, and up to 12x speedup of the time-to-solution with respect to a standard Bayesian Optimization procedure.

Efficient Performance Prediction for Apache Spark

Spark is a more efficient distributed big data processing framework following Hadoop. It provides users with more than 180 adjustable configuration parameters, and how to choose the optimal configuration automatically to make the Spark application run effectively is challenging. The key to address the above challenge is having the ability to predict the performance of Spark applications in different configurations. This paper proposes a new approach based on Adaboost, which can efficiently and accurately predict the performance of a given application with a given Spark configuration. In our approach, Adaboost is used to build a set of performance models at the stage-level for Spark. To minimize the overhead of the modeling, we use the classic projective sampling, a data mining technique that allows us to collect as few training samples as possible while meeting the accuracy requirements. We evaluate the proposed approach on six typical Spark benchmarks with five input datasets. The experimental results show that our approach is less than the previously proposed approach in prediction error and cost.

The Frequency-dependent Behavior of Subpulse Drifting. I. Carousel Geometry and Emission Heights of PSR B0031–07

Abstract The carousel model of pulsar emission attributes the phenomenon of subpulse drifting to a set of discrete sparks located very near the stellar surface rotating around the magnetic axis. Here, we investigate the subpulse drifting behavior of PSR B0031−07 in the context of the carousel model. We show that B0031−07's three drift modes (A, B, and C) can be understood in terms of a single carousel rotation rate if the number of sparks is allowed to change by an integral number, and where the different drift rates are due to (first-order) aliasing effects. This also results in harmonically related values for P 3 (the time it takes a subpulse to reappear at the same pulse phase), which we confirm for B0031−07. A representative solution has [n A, n B, n C] = [15, 14, 13] sparks and a carousel rotation period of P 4 = 16.4 P 1. We also investigate the frequency dependence of B0031−07's subpulse behavior. We extend the carousel model to include the dual effects of aberration and retardation, including the time it takes the information about the surface spark configuration to travel from the surface up to the emission point. Assuming these effects dominate at B0031−07's emission heights, we derive conservative emission height differences of ≲2000 km for mode A and ≲1000 km for modes B and C as seen between 185 and 610 MHz. This new method of measuring emission heights is independent of others that involve average profile components or the polarization position angle curve, and thus provides a potentially strong test of the carousel model.

Towards machine learning-based self-tuning of Hadoop-Spark system

Apache Spark is an open source distributed platform which uses the concept of distributed memory for processing big data. Spark has more than 180 predominant configuration parameter. Configuration settings directly control the efficiency of Apache spark while processing big data, to get the best outcome yet a challenging task as it has many configuration parameters. Currently, these predominant parameters are tuned manually by trial and error. To overcome this manual tuning problem in this paper proposed and developed a self-tuning approach using machine learning. This approach can tune the parameter value when it’s required. The approach was implemented on Dell server and experiment was done on five different sizes of the dataset and parameter. A comparison is provided to highlight the experimented result of the proposed approach with default Spark configuration system. The results demonstrate that the execution is speeded-up by about 33% (on an average) compared to the default configuration.

Test bench for studying the effects of water injection inside an internal combustion engine

Under the increasingly drastic conditions imposed on motor vehicles, in terms of pollution, manufacturers and researchers are looking for solutions to reduce pollutant emissions, especially emissions of nitrogen oxides (NOx) dissipated in the atmosphere. The problem of polluting emissions in the field of transport is becoming increasingly stringent, the pollution rules are becoming stricter, so there is a general interest in finding solutions to reduce pollutant emissions and fuel consumption. A possible solution to this problem can be achieved by direct or indirect water injection (through the valve port) inside the cylinder of an internal combustion engine. In this paperwork we present the test bench and we analyzed the effect of water injection inside a cylinder of an internal combustion engine. The water injection is made upstream of the throttle valve by single injector mounted on the intake. The injection timing and amount of water injected inside the engine have been controlled by CompactRIO hardware device and in-house algorithm implemented in LabView. The engine used for measurements is a 1.6 liter engine, manufactured by Daewoo Motor Company and for the moment it’s running on wasted spark configuration. So in order to have water particles in all the cylinders, the water is injected two times at each rotation of the crankshaft. The main goal of this research was decreasing of nitrogen oxides (NOx) by reducing the combustion chamber temperature.

Configuring in-memory cluster computing using random forest

Recently, in-memory cluster computing (IMC) gains momentum because it accelerates traditional on-disk cluster computing (ODC) up to several tens of times for iterative and interaction applications. The most popular IMC framework is Spark and it has more than 100 configuration parameters. However, it is unclear how significantly these parameters affect the system performance because IMC is a quite new computing paradigm. Consequently, there is yet no study addressing how to optimally configure IMC frameworks.In this paper, we first investigate how significantly the configuration parameters affect the performance of Spark workloads. We find that the configuration caused performance variation can be as large as 20.7, indicating configuring Spark workloads is extremely important to their performance. However, manually configuring Spark workloads is notoriously difficult because there are so many configuration parameters which might interfere with each other in a complex way. To address this issue, we propose an approach to Automatically Configure Spark workloads, named ACS. It firstly constructs performance models as functions of Spark configuration parameters by using random forest which is an ensemble learning algorithm. Subsequently, ACS leverages genetic algorithm to search the optimum configuration by taking configurations and the corresponding performance predicted by the performance models as inputs.We employ six Spark programs, each with five input data sets to evaluate the performance improvements. The results show that ACS speeds up the 30 program-input pairs by a factor of 2.2× on average and up to 8.2×. In addition, the performance improvements obtained by ACS increase along with the increments of the input data set sizes of Spark workloads, which is a nice property for big data analytics.

Evaluation of the Effects of a Twin Spark Ignition System on Combustion Stability of a High Performance PFI Engine

The continuous demand for high performances and low emissions engines leads the engine manufactures to set the operating range of combustion devices near to their stability limit. Combustion stability is closely related to the formation of the first ignition kernel: an effective way of lowering Cycle-by-Cycle Variation (CCV) is to enhance the start of combustion by means of multiple sparks. A Ducati engine was equipped with a Twin Spark ignition system and a consistent improvement in combustion stability arised for both part load and full load conditions.At part load a sensible reduction of cycle-by-cycle variability of indicated mean effective pressure was found, while at full load condition the twin spark configuration showed an increase of power, but with higher knocking tendency. The aim of this work is to better understand the root causes of the increased level of knock and to make a critical evaluation of most used knock indexes, by means of an accurate analysis of the experimental and simulated pressure signals.The numerical methodology based on a perturbation of the initial kernel by a statistical evaluation of mixture condition at ignition location. A lagrangian ignition model developed at University of Bologna was used, here modified to take into account the statistical distribution of mixture around the spark plugs. The RANS simulations proved to be accurate in representing all the main information related to combustion efficiency and knocking events.

Large-Eddy simulation analysis of spark configuration effect on cycle-to-cycle variability of combustion and knock

Cycle-to-cycle variability is numerically simulated for high-speed, full-load operation of a turbocharged gasoline direct injection engine. Large-Eddy simulation is adopted to replicate the fluctuations of the flow field affecting the turbulent combustion. Experimental data were provided at knock onset, and large-Eddy simulation was validated for the same condition. In the original engine configuration, the spark plug is displaced toward the exhaust side, while the electrodes orientation is arbitrary. A 90° rotation is imposed to evaluate the effects of the aerodynamic obstruction caused by the electrode with respect to the flow field and the flame kernel growth. A second speculative analysis is performed modifying the position of the spark plug. The electrodes are shifted 2mm toward the intake side since this variation is compatible with the cylinder head layout. For both variations in orientation and position, the effects on the flow field around the spark plug are investigated. Statistical analysis is carried out on early flame kernel formation and knock tendency. The results highlight that the orientation of the electrodes affects the flow field for each cycle but plays a negligible role on the statistical cyclic variability, indirectly justifying the lack of an imposed orientation. As for the spark plug position, the numerical analysis indicate that the shifting of the electrodes toward the intake side slightly improves the knock limit mainly because of a reduction in in-cylinder peak pressure. In general, it is inferred that improvements may be achieved only through a simultaneous modification of the fuel jet orientation and phasing.

Au-TiO2 Nanoscale Heterodimers Synthesis from an Ambient Spark Discharge for Efficient Photocatalytic and Photothermal Activity

Ultrafine Au particles incorporating TiO2 heterodimers were synthesized using an ambient heterogeneous spark discharge and the resultant materials were employed both in oxidizing photocatalytically CO gas and killing photothermally cancerous cells. Ti-Au spark configuration was employed to vaporize Ti and Au components into an airflow and finally ultrafine Au particles (∼2 nm in lateral dimension) were incorporated with TiO2 nanoparticles in the form of Au-TiO2 heterodimers (∼38 nm in lateral dimension) with enhanced photocatalytic (in CO oxidation) and photothermal activity (in cancerous cell killing) under visible light. We propose that the localized surface plasmon resonance of ultrafine Au particles on TiO2 supports, induced by the visible light, would promote the adsorption-oxidation of CO and photothermal killing of HeLa cells. The present strategy may be suitable to fabricate other Au-metal oxide nanocomposites for catalytic and biomedical applications.

Gas-Phase Self-Assembly of Highly Ordered Titania@Graphene Nanoflakes for Enhancement in Photocatalytic Activity

The gas-phase self-assembly of reduced graphene oxide (rGO) nanoflakes with highly ordered ultrafine titania (TiO2) particles was performed and the resultant hybrid material displayed an enhanced photocatalytic performance, both in producing hydrogen and in degrading dyes. Freshly synthesized TiO2 nanoparticles (∼35 nm in equivalent mobility diameter) were quantitatively incorporated with nanoscale rGO (∼36 nm in equivalent mobility diameter) in the form of TiO2/rGO hybrid nanoflakes (∼31 nm in equivalent mobility diameter). The TiO2/rGO hybrid flakes were finally employed to evaluate its photocatalytic activity, and it was found that the ability to achieve hydrogen production and dye degradation was greater than that of a hybridized material from commercial p25-TiO2 and large rGO. This gas-phase self-assembly also enhanced the photocatalytic activity by applying different spark configurations to prepare ZnO, Au, or Ag particles incorporated with rGO nanoflakes.

Ambient spark generation to synthesize carbon-encapsulated metal nanoparticles in continuous aerosol manner

We report the use of spark generation in an inert gas atmosphere to synthesize carbon-encapsulated metal nanoparticles (CEMNs) in a continuous aerosol manner using a metal (nickel, cobalt, iron)-graphite carbon electrode configuration without the use of a vacuum. The spark-generated particles consisted of CEMNs and carbonaceous aggregated debris. The outer layer of the CEMNs showed parallel fringes (ordered graphitic nanostructures) while the debris consisted of disordered nanostructures. Electron and X-ray diffraction showed that both metal and graphite in the CEMNs were the pure elements except for iron-carbon, which contained a carbide phase. Based on the order of the activation energies for carbon diffusion into a metal: iron-carbon (10.5-16.5 kcal mol(-1)) < cobalt-carbon (34.7 kcal mol(-1)) approximately nickel-carbon (33.0-34.8 kcal mol(-1)), it was concluded that carbide particles form more easily from elemental iron than nickel or cobalt. The metal-to-carbon mass fractions of the spark-generated particles from nickel (anode)-carbon (cathode), cobalt-carbon, and iron-carbon spark configurations were 18.7, 28.3, and 11.2%, respectively, while the mass fractions for the configurations of metal (cathode)-carbon (anode) were 6.4, 9.1, and 4.3%, respectively. Similarly, the yield of CEMNs from the metal (anode)-carbon (cathode) electrodes was higher (54, 61, and 53%) than that of metal (cathode)-carbon (anode) electrodes (18, 30, and 18%).

A low energy tunable pulsed X-ray source based on the pseudospark electron beam

A low energy pulsed X-ray source with 17 J electrical input energy has been constructed based on the vacuum spark configuration triggered by the pseudospark electron beam. With such a low input electrical energy, the frequently studied X-ray emitting vacuum spark plasma is not obtained. Instead, this configuration gives rise to a point-like electron beam-target pulsed X-ray source at the tip of the anode. Emission of X-ray is observed during the pre-breakdown phase. The X-ray spectrum is dominated by the characteristic line emission of the anode material. The total X-ray energy emission from the source can be estimated by assuming it to be a monochromatic point source. It is found to have an X-ray production efficiency of 0.1%, giving an average X-ray dosage of 18 mGy per pulse. X-ray radiography of small biological sample is demonstrated to illustrate a potential application of the present X-ray source.