Parametric Analysis with OpenStudio
The previous chapter introduced the concept of OpenStudio Measures and how they can be applied individually and in combination to a Model to create and compare different Design Alternatives. While an improvement from modifying models by hand, generating r
- PDF / 4,377,919 Bytes
- 48 Pages / 439.37 x 666.142 pts Page_size
- 53 Downloads / 401 Views
Parametric Analysis with OpenStudio
7.1 Introduction to Parametric Analysis The previous chapter introduced the concept of OpenStudio Measures and how they can be applied individually and in combination to a Model to create and compare different Design Alternatives. While an improvement from modifying models by hand, generating results, and comparing them; the manual analysis workflow is still labor intensive, non-scalable, and will not necessarily yield the best solution for a given problem. In this chapter, we will discuss how OpenStudio enables automated creation and search of large building parameter spaces. We’ll also look at how these same approaches may be used to “tune” models of existing buildings to best match measured energy consumption data.
7.2 OpenStudio Server In the previous chapter, we alluded to a “server” that ran on a user’s computer to manage multiple simulations in PAT’s manual mode. To understand and utilize OpenStudio for parametric analysis, it is important to have a better understanding of what OpenStudio Server is and (to a limited degree) how it works. Figure 7.1 illustrates the basic relationship between PAT, the “OpenStudio Server,” and the workers who create and simulate the individual data points. OpenStudio Server has three primary components: 1. Web Interface – A minimal interface that allows for user interaction with analysis projects and the contents of the Results Database, 2. Results Database – A database used to store high level simulation results for a project, and 3. R – An open source platform for statistical computing and analysis.1 https://www.r-project.org/.
1
© Springer International Publishing AG, part of Springer Nature 2018 L. Brackney et al., Building Energy Modeling with OpenStudio, https://doi.org/10.1007/978-3-319-77809-9_7
215
216
7 Parametric Analysis with OpenStudio
Fig. 7.1 PAT, OpenStudio Server, and Workers
OpenStudio Workers are a fourth component in the overall architecture, separate from but integral to the server. Each worker node is an independent computing instance, configured with OpenStudio, EnergyPlus, and supporting software. When provided with a seed Model, weather file, and one or more Measures; a worker has everything it needs to create, simulate, and post-process results for a given Design Alternative. While workers generate all of the output available in a typical simulation run, file size and storage capacity generally dictate that only a subset of that data be returned to the Results Database for subsequent use. We’ll discuss shortly how Reporting Measures, along with PAT’s Outputs ( ) Tab, are used to specify which values are stored. R does most of the “heavy lifting” for any OpenStudio Server-based analysis, defining individual data points to be simulated. The fundamental difference between the “mini server” used locally for manual analyses and a full OpenStudio Server implementation used in algorithmic mode is the inclusion of R and supporting files. The manner in which R defines data points, reviews results, prescribes
Data Loading...
