Performance Prediction Techniques Research Articles

Data Analytics Techniques for Performance Prediction of Steamflooding in Naturally Fractured Carbonate Reservoirs

Thermal oil recovery techniques, including steam processes, account for more than 80% of the current global heavy oil, extra heavy oil, and bitumen production. Evaluation of Naturally Fractured Carbonate Reservoirs (NFCRs) for thermal heavy oil recovery using field pilot tests and exhaustive numerical and analytical modeling is expensive, complex, and personnel-intensive. Robust statistical models have not yet been proposed to predict cumulative steam to oil ratio (CSOR) and recovery factor (RF) during steamflooding in NFCRs as strong process performance indicators. In this paper, new statistical based techniques were developed using multivariable regression analysis for quick estimation of CSOR and RF in NFCRs subjected to steamflooding. The proposed data based models include vital parameters such as in situ fluid and reservoir properties. The data used are taken from experimental studies and rare field trials of vertical well steamflooding pilots in heavy oil NFCRs reported in the literature. The models show an average error of <6% for the worst cases and contain fewer empirical constants compared with existing correlations developed originally for oil sands. The interactions between the parameters were considered indicating that the initial oil saturation and oil viscosity are the most important predictive factors. The proposed models were successfully predicted CSOR and RF for two heavy oil NFCRs. Results of this study can be used for feasibility assessment of steamflooding in NFCRs

Capacity Management as a Service for Enterprise Standard Software

Capacity management approaches optimize component utilization from a strong technical perspective. In fact, the quality of involved services is considered implicitly by linking it to resource capacity values. This practice hinders to evaluate design alternatives with respect to given service levels that are expressed in user-centric metrics such as the mean response time for a business transaction. We argue that utilized historical workload traces often contain a variety of performance-related information that allows for the integration of performance prediction techniques through machine learning. Since enterprise applications excessively make use of standard software that is shipped by large software vendors to a wide range of customers, standardized prediction models can be trained and provisioned as part of a capacity management service which we propose in this article. Therefore, we integrate knowledge discovery activities into well-known capacity planning steps, which we adapt to the special characteristics of enterprise applications. Using a real-world example, we demonstrate how prediction models that were trained on a large scale of monitoring data enable cost-efficient measurement-based prediction techniques to be used in early design and redesign phases of planned or running applications. Finally, based on the trained model, we demonstrate how to simulate and analyze future workload scenarios. Using a Pareto approach, we were able to identify cost-effective design alternatives for an enterprise application whose capacity is being managed.

Model-Based Resource Utilization and Performance Risk Prediction using Machine Learning Techniques

The growing complexity of modern software systems makes the performance prediction a challenging activity. Many drawbacks incurred by using the traditional performance prediction techniques such as time consuming and inability to surround all software system when large scaled. To contribute to solving these problems, we adopt a model-based approach for resource utilization and performance risk prediction. Firstly, we model the software system into annotated UML diagrams. Secondly, performance model is derived from UML diagrams in order to be evaluated. Thirdly, we generate performance and resource utilization training dataset by changing workload. Finally, when new instances are applied we can predict resource utilization and performance risk by using machine learning techniques. The approach will be used to enhance work of human experts and improve efficiency of software system performance prediction. In this paper, we illustrate the approach on a case study. A performance training dataset has been generated, and three machine learning techniques are applied to predict resource utilization and performance risk level. Our approach shows prediction accuracy within 68.9 % to 93.1 %.

A quantitative roofline model for GPU kernel performance estimation using micro-benchmarks and hardware metric profiling

Typically, the execution time of a kernel on a GPU is a difficult to predict measure as it depends on a wide range of factors. Performance can be limited by either memory transfer, compute throughput or other latencies. In this paper, we improve on the roofline model following a quantitative approach and present a completely automated GPU performance prediction technique. In this respect this model utilizes micro-benchmarking and profiling in a “black box” fashion as no inspection of source/binary code is required. The proposed model combines parameters in order to characterize the performance limiting factor and to estimate execution time. In addition, we propose the quadrant-split visual representation, which captures the characteristics of multiple processors in relation to a particular kernel. We performed experiments on stencil computation (red/black SOR), SGEMM and a total of 28 kernels of the Rodinia benchmark suite, using six CUDA GPUs and we showed an absolute error in predictions of 27.66% in the average case. Furthermore, the performance model was also examined on an AMD GPU through the HIP programming environment. Prediction errors were comparable despite the significant architectural differences between different vendor GPUs.

Numerical and experimental investigation of thermoelectric cooling in down-hole measuring tools; a case study

Use of Peltier cooling in down-hole seismic tooling has been restricted by the performance of such devices at elevated temperatures. Present paper analyses the performance of Peltier cooling in temperatures suited for down-hole measuring equipment using measurements, predicted manufacturer data and computational fluid dynamic analysis. Peltier performance prediction techniques is presented with measurements. Validity of the extrapolation of thermoelectric cooling performance at elevated temperatures has been tested using computational models for thermoelectric cooling device. This method has been used to model cooling characteristics of a prototype downhole tool and the computational technique used has been proven valid.

Low complexity link level performance prediction for SIMO systems

In this paper a low-cost link level performance prediction technique is proposed for a single input and multiple output system. Receiver link level abstraction is used in system level simulations of large networks in order to reduce their complexity. Usually, a single lookup table is employed in link level abstraction to predict a receiver's performance under different channel conditions. In the presented work, the mean frame error rate curve of different diverse channels is proposed as the reference for performance prediction in single input multiple output systems. Its generation involves compression of the received code word into a single quality measure based on the postdetection signal to noise ratio values using nonlinear capacity, exponential, and mutual mapping functions. The overall performance difference between simulated and predicted frame error rates shows that the proposed scheme gives very good performance approximations under different modulation and coding schemes, clearly outperforming the classical line of sight channel lookup table.

Predicting Student Performance Using Personalized Analytics

To help solve the ongoing problem of student retention, new expected performance-prediction techniques are needed to facilitate degree planning and determine who might be at risk of failing or dropping a class. Personalized multiregression and matrix factorization approaches based on recommender systems, initially developed for e-commerce applications, accurately forecast students' grades in future courses as well as on in-class assessments.

極細繊維材の吸音率予測手法の開発

This report deals with sound absorption coefficient prediction technique of ultrafine fiber. As the sound absorption material used for the interior noise reduction of the automobile, the fiber with a diameter of several μm is considered. When a fiber diameter becomes small, usually a sound absorption coefficient increases. But predictive precision was bad by the conventional performance prediction technique. I developed the technique to predict a sound absorption coefficient from the fiber diameter and the fiber density and the material thickness and the material density. The experiment expression of relations was drawn from the experimental results about the flow resistivity and the thermal characteristic length. And I introduce the technique used Limp frame model to predict a sound absorption coefficient for the ultrafine fiber. There diameters are 1-4μm. This paper shows comparison of the experimental results and the calculation results, and this technique is useful.

A new method for reliable performance prediction of multi-stage industrial centrifugal compressors based on stage stacking technique: Part I – existing models evaluation

The growing acceptability of the natural gas as an energy source yields to place a higher research concentration on improving the centrifugal compressor design and performance prediction techniques. This is the first part of a two-part study that aims to introduce a new integrated method for reliable performance prediction of multi-stage industrial centrifugal compressor based on stage stacking technique. In this part, two most common existing models are adapted to be valid for industrial centrifugal compressors and the predicted characteristics are evaluated against measured data. Based on the conducted evaluation, a new model will be derived in the second part which incorporates the advantages of both approaches with more accurate prediction. Lüdtke method predicts the compressor design efficiency based on the correction of the derived experimental set of data. This approach has been applied for a single stage centrifugal compressor at a single speed and flow values. Recently, Casey–Robinson method uses a new set of algebraic equations to obtain the compressor curve with less dependency on the geometrical features. This model has been tested with a single stage turbocharger compressor where the performance curve can be derived without considering the mechanical stages. This approximation will introduce a more significant impact when this method is used for large industrial compressors. Therefore, these models will be improved to be suitable for multi-industrial centrifugal compressors and their prediction capability to estimate the performance map will be examined in this paper.

Neighbor Selection and Weighting in User-Based Collaborative Filtering

User-based collaborative filtering systems suggest interesting items to a user relying on similar-minded people called neighbors. The selection and weighting of these neighbors characterize the different recommendation approaches. While standard strategies perform a neighbor selection based on user similarities, trust-aware recommendation algorithms rely on other aspects indicative of user trust and reliability. In this article we restate the trust-aware recommendation problem, generalizing it in terms of performance prediction techniques, whose goal is to predict the performance of an information retrieval system in response to a particular query. We investigate how to adopt the preceding generalization to define a unified framework where we conduct an objective analysis of the effectiveness (predictive power) of neighbor scoring functions. The proposed framework enables discriminating whether recommendation performance improvements are caused by the used neighbor scoring functions or by the ways these functions are used in the recommendation computation. We evaluated our approach with several state-of-the-art and novel neighbor scoring functions on three publicly available datasets. By empirically comparing four neighbor quality metrics and thirteen performance predictors, we found strong predictive power for some of the predictors with respect to certain metrics. This result was then validated by checking the final performance of recommendation strategies where predictors are used for selecting and/or weighting user neighbors. As a result, we have found that, by measuring the predictive power of neighbor performance predictors, we are able to anticipate which predictors are going to perform better in neighbor-scoring-powered versions of a user-based collaborative filtering algorithm.

CFD Assisted Prediction of Hydrodynamic Parameters for Regenerator of Cryocooler

Computational Fluid Dynamics (CFD) is the emerging technique for analysis and performance prediction of a pulse tube refrigerator. The regenerator plays an important role in any cryorefrigerator, and hence to predict the realistic performance, proper boundary conditions are required in CFD tool. The modeling of regenerator requires appropriate definition and knowledge about hydrodynamic parameters viz. viscous resistance, inertial resistance and porosity. The present work report, CFD assisted prediction of hydrodynamic parameters for some widely used regenerator wire mesh structures selected on the basis of different porosity, which can predict the same pressure drop and friction factor obtained from experimental correlation equation. A unique value of viscous and inertial resistance is obtained iteratively for each mesh screen structures over a range of mass flow rate. These data can be readily used as boundary conditions for CFD simulations of the test section.

A new screening tool for evaluation of steamflooding performance in Naturally Fractured Carbonate Reservoirs

Appropriate production method selection for Viscous Oil (e.g., Heavy Oil, Extra Heavy Oil, and Bitumen) Naturally Fractured Carbonate Reservoirs (VO NFCRs) mostly depends on the quality of the fluid and reservoir properties. Selection of a particular production method for a reservoir is generally evaluated through an exhaustive experimental, field pilot, and mathematical modeling approach. In the absence of robust and quick predictive tools, using connectionist techniques for performance prediction of a particular production method can be a valuable asset. In this study, a new screening tool is developed based on Artificial Neural Networks (ANNs) optimized with Particle Swarm Optimization (PSO) to assess the performance of steamflooding in VO NFCRs. As expected, Recovery Factor (RF) and Cumulative Steam to Oil Ratio (CSOR) during steamflooding are highly affected by the magnitudes of oil saturation and viscosity. The developed PSO–ANN model, conventional ANN and statistical correlations were examined using real data. Comparison of the predictions and real data implies the superiority of the proposed PSO–ANN model with an absolute average error percentage<6.5%, a determination coefficient (R2)>0.98, and Mean Squared Error (MSE)<0.06, in contrast with conventional ANN model and empirical correlations for prediction of RF and CSOR. This indicates a great potential for application of hybrid PSO–ANN models to screen Viscous Oil carbonate reservoirs for steamflooding.

209 可動性を考慮したガス遮断器内の三次元ガス挙動数値解析(OS3-(2)オーガナイズドセッション《熱・流体・エネルギー変換工学の進展》)

Gas circuit breaker is the very important safety device that instantaneously intercepts the circuit and protects the instrument when an abnormal current like lightning flows to electric power transmission line of high voltage transformation installation. Recently The miniaturization needs of the instrument builds up as a high glasshouse effect of SF6 is feared and the quantity consumed reduction is requested. In this study, we analyzed the behavior of the high temperature gas in consideration of three dimensions in gas circuit breaker,using CRIMSON(Civa Riflned Multiphase SimulasiON). And we thought separation of fixed electrode and movable electrode to attempt the improvement of interception performance prediction technique.

CHARACTERISTICS OF COMBINED DRAINAGE IN THE HOUSE DRAIN OF THE DRAINAGE SYSTEM WITH SPECIAL FITTINGS AND THE APPLICATION OF A PREDICTION TECHNIQUE USING THE PIPE NET WORK MODEL

This study intends to propose a testing method for examining the influence of combined drainage in the house drain on the drainage performance of the stack drainage system which is used for housing complexes. The study also intends to propose a drainage performance prediction technique using the testing method, and subsequently discusses the effectiveness of the testing method and the prediction technique. Subsequent to the previous paper, this paper aims to clarify, using the drainage system with special fittings, the following three points: (1) the aeration property of components of the model, which was proposed in the previous paper for predicting the pressure in the house drain when combined drainage occurs, and evaluate the adaptability of the model, (2) using the model, the prediction of the drainage performance of the house drain which is connected to the stack drainage system and compare acquired values to the SHASE-S206 values, (3) the discussion on the vertical height of the horizontal fixture drain branch against combine drainage, i.e. the horizontal fixture drain branch installable at the lowest floor from the house drain. Based on the above points, this paper also looks at the characteristics of combined drainage in the house drain, which are different when the house drain is on the drainage system with special fittings and when it is on the drainage system with JIS-DT fittings.

Classifier performance estimation under the constraint of a finite sample size: Resampling schemes applied to neural network classifiers

In a practical classifier design problem the sample size is limited, and the available finite sample needs to be used both to design a classifier and to predict the classifier's performance for the true population. Since a larger sample is more representative of the population, it is advantageous to design the classifier with all the available cases, and to use a resampling technique for performance prediction. We conducted a Monte Carlo simulation study to compare the ability of different resampling techniques in predicting the performance of a neural network (NN) classifier designed with the available sample. We used the area under the receiver operating characteristic curve as the performance index for the NN classifier. We investigated resampling techniques based on the cross-validation, the leave-one-out method, and three different types of bootstrapping, namely, the ordinary, .632, and .632+ bootstrap. Our results indicated that, under the study conditions, there can be a large difference in the accuracy of the prediction obtained from different resampling methods, especially when the feature space dimensionality is relatively large and the sample size is small. Although this investigation is performed under some specific conditions, it reveals important trends for the problem of classifier performance prediction under the constraint of a limited data set.

Improving a one-dimensional centrifugal compressor performance prediction method

A one-dimensional centrifugal compressor performance prediction technique that has been available for some time is updated as a result of extracting the component performance from three-dimensional computational fluid dynamic (CFD) analyses. Confidence in the CFD results is provided by comparison of overall performance for one of the compressor examples. The extracted impeller characteristic is compared with the original impeller loss model, and this indicated that some improvement was desirable. The position of least impeller loss was determined using a traditional axial compressor cascade method, and suitable algebraic expressions were derived to match the CFD data. The merit of the approach lies with the relative ease that CFD component performance currently can be achieved and adjusting one-dimensional methods to agree with the CFD-derived models.

Investigation of the scalability of dynamic wavelength-routed optical networks [Invited

Feature Issue on Optical Interconnection Networks (OIN). We describe results of the scalability analysis for dynamic wavelength-routed optical networks with end-to-end lightpath assignment and central network control with electronic scheduling and processing of lightpath requests. We investigate the effect of the algorithm complexity in both the scheduling and the dynamic routing and wavelength assignment (DRWA) of lightpath requests. Scheduling theory and static performance-prediction techniques were applied to define the bounds on the electronic processing time of requests, and hence the maximum number of nodes supported by a centralized dynamic optical network for given blocking probability, latency, and network diameter. Scalability analysis results show that medium-sized centralized networks (~50 nodes) can be supported when these networks are reconfigured on a burst-by-burst basis. In addition, we found that real topologies showed a complex trade-off between the request processing time, blocking probability, and resource requirements. These findings can be used to determine the optimum combination of scheduling/DRWA algorithm, showing that the fastest DRWA algorithm does not necessarily lead to the minimum blocking probability and maximum scalability but that a careful consideration of both blocking and processing speed is required. The results are applicable both to dynamic network architectures with centralized request processing such as wavelength-routed optical networks and to the design of advanced optical switching matrices and routers.

Revolving-field polygon technique for performance prediction of single-phase induction motors

This paper presents a new analytical technique for improving the performance prediction of single-phase induction motors, especially capacitor motors. The technique uses the split-phase motor electrical equivalent circuit analysis together with electrical and magnetic parameters whose variation is computed from the equivalent balanced polyphase motor, so that the same magnetic circuit analysis can be used for both. (The term split-phase is used to cover motors operating from a single-phase supply but with the phase windings split into two orthogonal windings, one of which may have a capacitor in series with it during running or starting.) The technique accounts for the elliptical envelope of the magnetizing field vector and results in improved precision, since the three-phase electromagnetic model is considered to be more precise than the normal split-phase motor analysis. An important result is the computation of vector polygons of flux density for each section of the magnetic circuit, providing a better basis for core loss prediction. The double-frequency torque ripple is also obtained from the stator magnetomotive force and flux-density polygons. Three different electrical equivalent circuit methods for the split-phase motor (based respectively on the cross-field theory, forward- and backward-revolving fields, and symmetrical components) are evaluated to determine the method best suited for incorporating the variation of the circuit parameters from the polyphase magnetic circuit analysis, and it is discussed how the core losses can be included in these circuits to obtain the best overall performance prediction.

Neural network techniques for financial performance prediction: integrating fundamental and technical analysis

This research project investigates the ability of neural networks, specifically, the backpropagation algorithm, to integrate fundamental and technical analysis for financial performance prediction. The predictor attributes include 16 financial statement variables and 11 macroeconomic variables. The rate of return on common shareholders' equity is used as the to-be-predicted variable. Financial data of 364 S&P companies are extracted from the CompuStat database, and macroeconomic variables are extracted from the Citibase database for the study period of 1985–1995. Used as predictors in Experiments 1, 2, and 3 are the 1 year's, the 2 years', and the 3 years' financial data, respectively. Experiment 4 has 3 years' financial data and macroeconomic data as predictors. Moreover, in order to compensate for data noise and parameter misspecification as well as to reveal prediction logic and procedure, we apply a rule extraction technique to convert the connection weights from trained neural networks to symbolic classification rules. The performance of neural networks is compared with the average return from the top one-third returns in the market (maximum benchmark) that approximates the return from perfect information as well as with the overall market average return (minimum benchmark) that approximates the return from highly diversified portfolios. Paired t tests are carried out to calculate the statistical significance of mean differences. Experimental results indicate that neural networks using 1 year's or multiple years' financial data consistently and significantly outperform the minimum benchmark, but not the maximum benchmark. As for neural networks with both financial and macroeconomic predictors, they do not outperform the minimum or maximum benchmark in this study. The experimental results also show that the average return of 0.25398 from extracted rules is the only compatible result to the maximum benchmark of 0.2786. Consequentially, we demonstrate rule extraction as a postprocessing technique for improving prediction accuracy and for explaining the prediction logic to financial decision makers.

Environment for multiprocessor simulator development

AbstractTo develop multiprocessor systems for continuing the precipitous increase in performance seen in recent years, it is important to include performance prediction at a preimplementation stage. Since software simulation, which is one such performance prediction technique, has a high degree of flexibility, it is used in diverse situations. To simulate a multiprocessor system, a simulator is implemented for a specific architecture. However, this method has accompanying implementation costs for building a simulator for each different target architecture. In this paper, the authors propose the multiprocessor simulator library ISIS as a multiprocessor simulator construction support system. ISIS is a collection of multiprocessor system internal function block simulators. Multiprocessor simulators having any kind of architecture can be constructed by combining function block simulators. Compared with conventional techniques, implementation costs can be reduced without directly impairing simulator runtime costs. Several multiprocessor simulators were implemented by using ISIS, and their execution speeds and implementation costs were evaluated. Sufficient runtime speeds were obtained for practical uses at extremely low implementation costs. ISIS has actually been used in research for on‐chip multiprocessor cache systems, and various kinds of knowledge have been obtained from the evaluation results of that research. © 2003 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 86(10): 74–84, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.10122