Building design is a multi-criteria decision-making process that typically involves numerous design parameters, such as geometry and envelope characteristics, uncertainty in internal loads, various HVAC system characteristics, utility rate structures, etc. Extensive simulation runs are required to complete the associated combinatorial parametric sensitivity analysis, which may not be realistic due to the sheer number of design possibilities, variables, and potential design criteria (response variables or objective functions). Additionally, it is impractical to use traditional optimization techniques to conduct a thorough search for optimal and sub-optimal solutions, even though in many cases, such solutions may end up being preferable options when financial effects, site-related constraints, or aesthetic preferences are considered.
The strategy used in this work is to classify the parameters according to their significance and degree of interaction, focusing more on those that have a bigger influence and interaction through statistical methods, graphical plots and machine learning algorithms. Naturally, the outcomes determined in this work are, of course, unique to this circumstance and area and should not be generalized to other building kinds and locations.
Refer to the Demonstration Videos to understand the complete methodology
Dowload detailed Executive Summary
Screening process using Morris Method (OAT – one at a time method) and Sobol method to identify the relevant parameters
The chart plots all the parameters initially in consideration. The importanct / impact of parameters increase as we move left to right on the chart and the interativity of parameters imcrease as we move from bottom to top of the chart. Therefore the parameter plotted on top right would have most influence on the objective function both in terms of first order impact and higher order interaction. The parameters plotted towards the bottom left corner of chart would be insignificant for further consideration / analysis. This chart has been plotted with EUI as the objective function. We can determine parameters based on other objective functions as well by plotting similar charts.
Exhaustive Latitude Analysis
The parameters with significant impact and interactions have been further analyzed and plotted on the chart below where designers can apply filters / constraints on various objective functions as well as parameters to make a well informed decision.
Interaction Chart
A parameter's importance is indicated by the size of the circles that are associated with it (i.e., the overall impact that parameter can have on an objective function while other parameters are left to vary), and the interaction between two parameters is indicated by the thickness of the line that connects them
Conclusions
The designers can explore a wide range of potential combinations based on the requirements of the building owner as this tool offers clear and better interpretation of the outcomes in addition to user-friendliness.
With much less computing time and resources, a high level of accuracy can be achieved in the results.
The effects of many design factors can be assessed with substantially fewer simulation runs, while also revealing information about parameter significance and parameter interaction.