Call (+1) 650-212-1212


The role of Sensitivity Analysis in Energy and Environmental Modeling

Sean Salleh 18 Jun 2013 Energy and environment

Energy and environmental models are often complex. Sensitivity analysis is a natural choice to improve understanding of such models and to see which input parameters have the most influence on the output. It also helps analysts to understand the inner workings of a model and refine it accordingly. Examples range from the design of buildings for energy conservation to the environmental impact of oil spills. By better identifying the sources of uncertainty, changes in data sampling methods may also be made in order to create a more robust model.

Diagram of Energy Model Image source: Lawrence Livermore National Laboratory

Sensitivity Analysis in Environmental Models

Environmental parameters, depending on context, include wind, humidity, cloud cover, rainfall, currents, tides, temperatures and solar exposition. Sensitivity analysis can help in picking the most influential parameters to be included in a model, while leaving aside those that are unimportant. Given the difficulty of estimating values for some environmental parameters, it is useful to understand whether or not time and effort should be spent in trying to integrate them into the model. By concentrating on fewer, but more relevant parameters, model size can be reduced, simulation times can be shortened and better output obtained.

Diagram of sun and earth interaction Image source:

Sensitivity Analysis in Energy Models

Energy and environment are typically linked. A model for energy will typically include output on factors such as carbon emission such as in the Belgrade Domestic Energy Model. This model uses sensitivity analysis to study differences in housing of different types and different ages. It also recognizes that complexity is still part of the model make-up because of the difficulty in assessing the behavior and interaction of a number of the input parameters. Other national energy models use physical laws, information on competition between energy sources, network effects (how the attractiveness of an energy is determined by the extent to which it is used), government policies and likely investment to model outcomes. Each factor can then be assessed for sensitivity as regards the final result.

Energy and Environmental Model Methods

Definitions of how models will be constructed in terms of information sampling and analysis of sensitivity and uncertainty (the overall measure of the degree of confidence possible in a model) vary. Possibilities include:

  • Sampling. Monte Carlo, Quasi-Monte Carlo (deterministic), and Latin Hypercube.
  • Sensitivity. Elementary effects/Morris method, ANOVA
  • Uncertainty. Standard deviation, variance, covariance

Data to be input into these models may come from datasets or from automated methods to extract from websites whose data is updated on a regular basis. While these models require training in the different methodologies used, they do not exclude other initiatives. Modeling and understanding the environment is a topic, albeit in simpler formats, that is now present in a number of environment teaching programs for children and teenagers.

Building planning diagram Image source: Buildipedia

The Results of Sensitivity Analysis

By refining the model and making it as robust as possible, energy and environmental trends can be calculated and mapped. The recent plan for reinforcing resilience for New York City, following Hurricane Sandy and including coastal protection, transport and communications, was the result of models of climatic conditions and trends using sensitivity analysis. In other cases, the analysis allows critical questions to be asked about sustainable futures for types of transport, or the speed at which new policies can be enacted to appropriately change society’s use of energy resources and its impact on the environment.

» Back

Sean Salleh

Sean Salleh is a data scientist with experience in guiding marketing strategy from building marketing mix models, forecasting models, scenario planning models, and algorithms. He is passionate about consumer technologies and resource management. He has master's degrees in Operations Research from University of California Irvine and Mathematics from Northeastern University.

Leave a Comment