Sizing Allocation Research Articles

Task Partitioning and Load Balancing Strategy for Matrix Applications on Distributed System

In this paper, we present a load-balancing strategy (Adaptive Load Balancing strategy) for data parallel applications to balance the work load effectively on a distributed system. We study its impact on computation-hungry matrix multiplication application. The ALB strategy enhances the performance with features such as intelligent node selection, pre-task assignment, adaptive task sizing and buffer allocation, and load balancing. The ALB strategy exhibits reduced nodes idle time and inter process communication time, and improved speed up as compared to Run Time task Scheduling strategy.

The role of anthropometry in designing for sustainability

An understanding of human factors and ergonomics facilitates the design of artefacts, tasks and environments that fulfil their users’ physical and cognitive requirements. Research in these fields furthers the goal of efficiently accommodating the desired percentage of user populations through enhanced awareness and modelling of human variability. Design for sustainability (DfS) allows for these concepts to be leveraged in the broader context of designing to minimise negative impacts on the environment. This paper focuses on anthropometry and proposes three ways in which its consideration is relevant to DfS: reducing raw material consumption, increasing usage lifetimes and ethical human resource considerations. This is demonstrated through the application of anthropometry synthesis, virtual fitting, and sizing and adjustability allocation methods in the design of an industrial workstation seat for use in five distinct global populations. This work highlights the importance of and opportunities for using ergonomic design principles in DfS efforts. Practitioner Summary: This research demonstrates the relevance of some anthropometry-based ergonomics concepts to the field of design for sustainability. A global design case study leverages human variability considerations in furthering three sustainable design goals: reducing raw material consumption, increasing usage lifetimes and incorporating ethical human resource considerations in design.

Operational scheduling of refined products pipeline networks with simultaneous batch injections

Petroleum refined products are mostly sent from oil refineries to distribution depots by trunk pipelines. Pipeline networks usually involve multiple input and exit terminals, and even dual-purpose stations. Several pumping operations can be simultaneously performed at different sources. Most of the computational burden on the scheduling of multi-source pipeline networks comes from three operational tasks: pump sequencing, batch sizing, and batch allocation. Previous contributions applied discrete decomposition approaches performing such tasks through heuristic-based decisions. This paper introduces an MILP continuous formulation for the operational scheduling of unidirectional pipeline networks that allows simultaneous batch injections. The problem goal is to satisfy depot requirements at minimum total cost. The optimal schedule of pumping and delivery operations is established all at once. Results show that simultaneous batch injections lead to a better use of the pipeline transport capacity and a substantial reduction on the overall time needed to meet depot demands.

Variability-Driven Formulation for Simultaneous Gate Sizing and Postsilicon Tunability Allocation

Process variations cause design performance to become unpredictable in deep submicrometer technologies. Several statistical techniques (timing analysis, gate sizing, and buffer insertion) have been proposed to counter these variations during the optimization phase of the design flow to get a better timing yield. Another interesting approach to improve the timing yield is postsilicon-tunable (PST) clock tree. In this paper, we propose such an integrated framework that performs simultaneous statistical gate sizing in the presence of PST clock-tree buffers for minimizing binning yield loss (YL) and tunability costs by determining the ranges of delay tuning to be provided at each buffer. The simultaneous gate sizing and PST-buffer range determination problem is proved to be a convex-stochastic programming formulation under longest path-delay constraints and, hence, solved optimally. We further extend the formulation into a heuristic to additionally consider shortest path-delay constraints. We make experimental comparisons using nominal gate sizing followed by PST-buffer management using the work of Tsai as a base case. We take the solution obtained from this approach and perform the following: 1) sensitivity-based statistical gate sizing while retaining the PST clock tree and 2) simultaneous gate sizing and PST-buffer range determination as proposed in this paper. On an average, the base-case approach gave 23% timing YL, the sensitivity approach gave 15% YL, whereas our proposed algorithm gave only 4% YL.

Production Equations for Unsteady-State Construction Processes

The production equation called Little's law has been applied to construction data recently. However, Little's law was derived for steady-state conditions assuming constant input and output rates and long production runs. Production in construction is inherently temporary, and learning curves and environmental influences often render input and output rates unequal and nonlinear. Starting with a conservation of mass formulation, general equations for work-in-process and cycle time for unsteady-state conditions and limited run production are developed. The motivation behind these equations is to explain common trends in production variables seen on construction projects. Previous studies have shown that when output from a construction production system is drastically increased, a significant upward impact is also seen on cycle time and work-in-process, and this work provides underlying theory and equations to explain these trends. Cycle time and work-in-process equations are presented as functions of time and on average. Data from construction activities are used to show that unsteady-state conditions commonly occur. Reasonable simplifications of the general equations provide guidelines for buffer sizing and resource allocation decisions.

Use of analytical performance models for system sizing and resource allocation in interactive video-on-demand systems employing data sharing techniques

In designing cost-effective video-on-demand (VOD) servers, efficient resource management and proper system sizing are of great importance. In addition to large storage and I/O bandwidth requirements, support of interactive VCR functionality imposes additional resource requirements on the VOD system in terms of storage space, as well as disk and network bandwidth. Previous works have used data sharing techniques (such as batching, buffering, and adaptive piggybacking) to reduce the I/O demand on the storage server. However, such data sharing techniques complicate the provision of VCR functions and diminish the amount of benefit that can be obtained from data sharing techniques. The main contribution of this paper is a simple, yet powerful, analytical modeling approach which allows for analysis, system sizing, resource allocation, and parameter setting for a fairly general class of data sharing techniques which are used in conjunction with the providing of VCR-type functionality. Using this mathematical model, we can determine the proper amount of resources to be allocated for normal playback as well as for service of VCR functionality requests while satisfying predefined system performance requirements. To illustrate the usefulness of our model, we focus on a specific data sharing scheme which combines the use of batching, buffering, and adaptive piggybacking, as well as allows for the use of VCR functions. We show how to utilize this mathematical model for system sizing and resource allocation purposes.

Distributed evolutionary algorithms for simulation optimization

The optimization of such complex systems as manufacturing systems often necessitates the use of simulation. In this paper, the use of evolutionary algorithms is suggested for the optimization of simulation models. Several types of variables are taken into account. The reduction of computing cost is achieved through the parallelization of this method, which allows several simulation experiments to be run simultaneously. Emphasis is put on a distributed approach where several computers manage both their own local population of solutions and their own simulation experiments, exchanging solutions using a migration operator. After a first evaluation through a mathematical function with a known optimum, the benefits of this new approach are demonstrated through the example of a transport lot sizing and transporter allocation problem in a manufacturing flow shop system, which is solved using a distributed software implemented on a network of eight Sun workstations.

A Model for Fleet Sizing and Vehicle Allocation

There are important interactions between decisions on sizing a vehicle fleet and decisions on utilizing that fleet. A model is formulated to optimize both sets of decisions simultaneously under dynamic and uncertain conditions. A network approximation to this model is presented and a solution procedure proposed. The convergence behavior of the solution procedure and the potential consequences of using simpler models to solve the fleet sizing and vehicle allocation problem are illustrated using several hypothetical problems. We conclude that this model is useful in identifying good strategies for the sizing of vehicle fleets and for the allocation of empty vehicles.