A decomposition procedure for solving two-dimensional distributed energy system design problems. (5th May 2016)

• Record Type: Journal Article
• Authors: Haikarainen, Carl; Pettersson, Frank; Saxén, Henrik
• Is Part Of: Applied thermal engineering. Volume 100(2016:May)
• Page Start: 30
• Page End: 38
• Publication Date: 2016-05-05
• Subjects: Optimisation -- District heating -- Distributed energy systems -- Spatial decomposition; Heat engineering -- Periodicals; Heating -- Equipment and supplies -- Periodicals; Periodicals; 621.40205
• DOI: 10.1016/j.applthermaleng.2016.02.012
• Languages: English
• ISSNs: 1359-4311

Abstract

Highlights: Structural and operational optimisation of distributed energy systems is discussed. A decomposition method to speed up the optimisation is suggested. The performance of the procedure is evaluated with three test cases. Optimisation with and without decomposition is compared. Abstract: Careful structural and operational design of distributed energy systems such as district heating networks can play an important part in efforts to reduce the use of energy and related resources by human societies. Mathematical models applied to analyse and optimise such energy systems are constantly improving, but limitations of their use still exist. One such limit can be the number of nodes in a network representation the optimisation routines can handle without solution times growing unpractically long. This article presents an iterative decomposition procedure for reducing the problem sizes and thus speeding up calculations when optimising district energy networks. The procedure involves clustering network nodes into a smaller number of sectors, which are treated as individual nodes in simplified networks that are optimised. This process is repeated, gradually increasing the detail level of the simplified networks. Solving the simplified networks gives information used to remove options from the more detailed network representations. The procedure is demonstrated by solving test cases, optimising networks in terms of economy and CO2 emissions, and comparing the solution qualities with solutions from an original mixed-integer linear optimisation model without decomposition. The new method was found to be a promising alternative for tackling problems that are too complex to be fully solved by state-of-the-art techniques.