Fernando Alvarruiz

Combining Neural Networks and Genetic Algorithms to Predict and Reduce Diesel Engine Emissions

Diesel engines are fuel efficient which benefits the reduction of CO2 released to the atmosphere compared with gasoline engines, but still result in negative environmental impact related to their emissions. As new degrees of freedom are created, due to advances in technology, the complicated processes of emission formation are difficult to assess. This paper studies the feasibility of using artificial neural networks (ANNs) in combination with genetic algorithms (GAs) to optimize the diesel engine settings. The objective of the optimization was to find settings that complied with the increasingly stringent emission regulations while also maintaining, or even reducing the fuel consumption. A large database of stationary engine tests, covering a wide range of experimental conditions was used for this analysis. The ANNs were used as a simulation tool, receiving as inputs the engine operating parameters, and producing as outputs the resulting emission levels and fuel consumption. The ANN outputs were then used to evaluate the objective function of the optimization process, which was performed with a GA approach. The combination of ANN and GA for the optimization of two different engine operating conditions was analyzed and important reductions in emissions and fuel consumption were reached, while also keeping the computational times low

An economic and energy-aware analysis of the viability of outsourcing cluster computing to a cloud

This paper compares the total cost of ownership of a physical cluster with the cost of a virtual cloud-based cluster. For that purpose, cost models for both a physical cluster and a cluster on a cloud have been developed. The model for the physical cluster takes into account previous works and incorporates a more detailed study of the costs related to energy consumption and the usage of energy-saving strategies. The model for the cluster on a cloud considers pricing options offered by Amazon EC2, such as reserving instances on a long-term basis, and also considers using tools for powering nodes on and off on demand, in order to avoid the costs associated to keeping idle nodes running. Using these cost models, a comparison is made of physical clusters with cloud clusters of a similar size and performance. The results show that cloud clusters are an interesting option for start-ups and other organizations with a high degree of uncertainty with respect to the computational requirements, while physical clusters are still more economically viable for organizations with a high usage rate.

An Energy Manager for High Performance Computer Clusters

This paper presents a general energy management system for HPC clusters and cloud infrastructures that powers off cluster nodes when they are not being used, and conversely powers them on when they are needed. This system can be integrated with different HPC cluster middleware, such as Batch-Queuing Systems or Cloud Management Systems, by using a set of connectors, and is also able to deal with different mechanisms for powering on and off the computing nodes (such as Wake-on-Lan, Power Device Units, Intelligent Platform Management Interface or other infrastructure-specific mechanisms). While some existing Batch-Queuing Systems provide energy saving mechanisms, other popular choices lack this feature. Cloud management middleware do not generally provide this feature out of the box, and incorporating it implies making modifications to the middleware. The advantage of our approach is that it can be integrated with different resource management middleware, without needing any modification of that middleware. The paper describes the successful integration of the system proposed with the popular Torque/PBS management system, and also with the OpenNebula open source cloud management tool. Two real use-cases are presented, involving two different HPC clusters. These use cases show significant energy/costs savings of 38% and 16%.

Infrastructure Deployment Over the Cloud

With the advent of cloud technologies the scientists have access to different cloud infrastructures in order to deploy all the virtual machines they need to perform the computations required in their research works. This paper describes a software architecture and a description language to simplify the creation of all the needed resources, and the elastic evolution of the computing infrastructure depending on the application requirements and some QoS features.

EC3: Elastic Cloud Computing Cluster

This paper introduces Elastic Cloud Computing Cluster (EC3), a tool that creates elastic virtual clusters on top of Infrastructure as a Service (IaaS) Clouds. The clusters are self-managed entities that scale out to a larger number of nodes on demand, up to a maximum size specified by the user. Whenever idle resources are detected, the clusters automatically scale in, according to some predefined policies, in order to cut down the costs in the case of using a public Cloud provider. This creates the illusion of a real cluster without requiring an investment beyond the actual usage. Two different case studies are presented to assess the effectiveness of an elastic virtual cluster. The results show that the usage of self-managed elastic clusters represents an important economic saving when compared both to physical clusters and to static virtual clusters deployed on an IaaS Cloud, with a reduced penalty in the elasticity management.

An energy management system for cluster infrastructures

This paper presents a general energy management system for High Performance Computing (HPC) clusters and cloud infrastructures that powers off cluster nodes when they are not being used, and conversely powers them on when they are needed. This system can be integrated with different HPC cluster middleware, such as Batch-Queuing Systems or Cloud Management Systems, and can also use different mechanisms for powering on and off the computing nodes. The presented system makes it possible to implement different energy-saving policies depending on the priorities and particularities of the cluster. It also provides a hook system to extend the functionality, and a sensor system in order to take into account environmental information. The paper describes the successful integration of the system proposed with some popular Batch-Queuing Systems, and also with some Cloud Management middlewares, presenting two real use-cases that show significant energy/costs savings of 27% and 17%.

Improving the performance of water distribution systems' simulation on multicore systems

Hydraulic solvers for the simulation of flows and pressures in water distribution systems (WDS) are used extensively, and their computational performance is key when considering optimization problems. This paper presents an approach to speedup the hydraulic solver using OpenMP with two efficient methods for WDS simulation. The paper identifies the different tasks carried out in the simulation, showing their contribution to the execution time, and selecting the target tasks for parallelization. After describing the algorithms for the selected tasks, parallel OpenMP versions are derived, with emphasis on the task of linear system update. Results are presented for four different large WDS models, showing considerable reduction in computing time.

Parallel Computing in Water Network Analysis and Optimization Processes

In this paper, a parallel computing based demonstrator for the simulation and control of water networks is presented. EPANET package has been the starting point for the demonstrator. The paper discusses the approach used in hydraulics simulation by means of the Gradient mefhod and in water quality simulation by means of the DVEM method. Finally, a key point in the project is the development of a methodology for leakage reduction by

Parallel Computing in Water Network Analysis and Leakage Minimization

nding the optimal pressure reducing valve settings.