Distributed and vectorized method of characteristics for fast transient simulations in water distribution systems

Riaño-Briceño, Gerardo; Sela, Lina; Hodges, Ben R.

doi:10.1111/mice.12709

Cited by 8 publications

(4 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It regards multiple decision points as a whole, and each individual can make the optimal selection of the game strategy of the next generation to be executed by himself. Moreover, the solutions of these participants are obtained after the solution to decide whether to continue or stop the operation and other processes, so as to maximize or minimize the overall decision-making, thus optimizing the overall benefit of the system and maximizing the global search ability [13][14]. In a distributed system, each task has to go through multiple decisions, and each scheme is composed of several non cooperative individuals, and also has an independent objective function.…”

Section: Hybrid Task Scheduling Algorithmmentioning

confidence: 99%

Hybrid Task Scheduling Algorithm Considering Game Model in Distributed System

2022

DPS

View full text Add to dashboard Cite

The distributed system has the characteristics of small scale, strong dynamics, and has an impact on both supply and demand. It plays a very important role in solving complex problems. In this paper, a mathematical model of hybrid task scheduling algorithm is established based on the consideration of profit maximization and uncertainty constraints in cooperative game among enterprises. Firstly, this paper studies the hybrid scheduling problem, then introduces the game model, and then studies the application of hybrid task scheduling algorithm. On this basis, the structure of distributed system is designed and the model is tested. Finally, the test results are shown. Under the optimization of hybrid task scheduling algorithm, the distributed system has improved the worst result and stability to a certain extent, and the improvement degree is great. The improvement effect of load balance degree is obvious and the stability is higher.

show abstract

Section: Hybrid Task Scheduling Algorithmmentioning

confidence: 99%

Hybrid Task Scheduling Algorithm Considering Game Model in Distributed System

2022

DPS

View full text Add to dashboard Cite

show abstract

“…Indeed, BIPquick's strategy of introducing phantom nodes to improve partition size balance might increase the overall computational cost of simulating a given system. Nevertheless, BIPquick is likely to prove valuable for water distribution models that use sub-discretization of links for transient analyses, e.g., Riaño-Briceño et al (2021), Duan et al (2020).…”

Section: Extended Applicationsmentioning

confidence: 99%

A Topological Approach to Partitioning Flow Networks for Parallel Simulation

Tiernan

Hodges

2022

J. Comput. Civ. Eng.

Self Cite

View full text Add to dashboard Cite

System partitioning for effective simulation of civil infrastructure flow networks on parallel processors is a non-trivial problem. Arbitrary partitioning focused only on balancing processor workload can lead to a large inter-processor communication burden that limits parallel speedup. Thus, there is a need for intelligent partitioning algorithms that balance the estimated computational load while minimizing the number of connections between partitions. Graph Theory provides widely-used partitioning methods, but these are applicable to networks with power-law connectivity and where the computational workload is proportional to the number of system nodes -conditions that do not hold for finite-volume solution of water drainage networks (e.g. river systems, stormwater drainage systems). This paper presents the novel "BIPquick" algorithm, which is shown to be an effective approach to identifying network partitions with reduced connectivity for systems that are Directed Acyclic Graphs (DAG) and have a physical limit on the number of connections per network node.Novel developments include (i) a "node-cut" approach that allows a partitioning workload function to be exactly balanced in systems where the computational work is proportional to the link length between nodes, (ii) a finite-pass approach to partitioning that ensures a partitioning solution in a known time, and (iii) a new connectivity scaling metric that allows simple evaluation and comparison of different partitioning results. The BIPquick model is tested on a large river network with up to 10,000 partitions. Tiernan, March 21, 2022 limited to junctions with less than 10 connecting links. Thus, practical approaches to the partitioning networks with limited junction connectivity are of interest to civil engineering.To further limit our scope, in finite-volume, unsteady flow models in a network the links ("edges" in Graph Theory) between nodes are subdivided into smaller elements of roughly uniform size such that the computational work scales on the link lengths -which contrasts with traditional link-node models (e.g., EPANET, SWMM) where the link length does not directly affect the workload. Hence, we focus herein on partitioning with link-dependent load balancing for gravity-drainage networks solved with finite-volume algorithms.As a simple example of the importance of both inter-processor communication and workload in partitioning, consider a network with N links of uniform length partitioned onto K processors where N ≫ K. The workload on an individual processor is (in this case) the number of links assigned to the processor. The overall communication burden is the number of nodes that connect to more than one processor. Figure 1a illustrates a worst-case condition for communication by using random sort, i.e. link-balance partitioning of an unsorted set. The random sort achieves workload balance, but at the cost of the inter-processor communication load scaling on N . In contrast, Fig. 1(b) illustrates, the reverse problem, a partitioning scheme that minimi...

show abstract

“…Secondly, localizing the origin of water-hammer waves necessitates a real-time high-frequency monitoring with widely distributed detectors, rarely deployed in WDNs. Thirdly, the computational cost of direct transient wave modeling in water distribution networks is very challenging either using the Method Of Characteristics (Wang et al, 2014;Nault et al, 2018;Meniconi et al, 2021;Moosavian and Lence, 2020;Riaño-Briceño et al, 2022), or finite volumes e.g (Pal et al, 2021;Zhang et al, 2021). Fourthly, an inverse method capable of identifying the origin of a reverberating water hammer wave over a large network, solving as many direct problems as tentative tries for possible origin, is obviously even more challenging.…”

Section: Introductionmentioning

confidence: 99%