In aroyal641/ebase: Energy Baseline and Prediction

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)

options(warn=-1)

ebase

The ebase package provides an ensemble of tools to model and visualize metered energy data. Its primary use-case is baseline modeling and savings calculations for net metered energy analysis, as it is used in M&V evaluation. However, the package also includes tools for other applications such as temperature profiling, loadpath cluster analysis and outlier detection.

Installation

You can install the development version from GitHub with:

# install.packages("devtools")
devtools::install_github("AndrewYRoyal/ebase")

Example

This is a example which shows how to use ebase to

1. Estimate a baseline model of metered consumption prior to a measure installation,
2. Use the baseline model to predict consumption before and after the installation, and
3. Use the predictions to calculate savings associated with the measure(s).

library(ebase)
library(magrittr)
library(data.table)

Data

The package includes sample two years hourly metered data for two simulated sites. Each meter is assigned a simulated measure installation date. In the example below, we calculate energy savings realized over the 365 days following the installation dates.

print(ebSample_sites)
print(ebSample_install_dates)
print(ebSample_deemed) # Engineering estimates for annual savings.
print(ebSample_hourly)

First, we format the hourly consumption data, dividing it into basleline, blackout and performance periods that signify the time intervals before, during and after the measure installations.

data_formatted = ebDataFormat(
  x = ebSample_hourly,
  install_dates = ebSample_install_dates,
  sites = ebSample_sites)

The resulting object can be used to return different versions of formatted data. This can be achieved using the ebStack() or ebList() methods. The ebStack() method creates a list of data.table objects, one for each period, where each table includes (stacks) readings from all meters. The ebList() method returns a list of lists, where each item contains a data.table object for the meter-period pairing.

names(ebStack(data_formatted))
names(ebList(data_formatted))
names(ebList(data_formatted)[['baseline']])
print(ebList(data_formatted)[['baseline']][['A']][, 1:5])

The ebList() formatting is useful when using lapply() to map a function over the entire set of meters, as we do when fitting models.

Modeling

To estimate model, we simply use the invoke the ebModel() function on the desired subset of formatted data. We can then use the resulting model to generate a predictions. The following code estimates a model using meter A's baseline period consumption (use), and then uses the model predict consumption (pUse) over all periods:

ebList(data_formatted)[['baseline']][['A']] %>%
  ebModel %>%
  predict(ebStack(data_formatted)[['A']])

Savings

To estimate total savings from measures, we must generate models and predictions for all three meters. This is best achieved using the lapply() function to map ebModel to all baseline meter data. The ebPredict() simplifies the prediction process for multiple meters. Lastly ebSummary and ebSavings() summarize the model predictions and calculate savings:

ebList(data_formatted)[['baseline']] %>%
  lapply(ebModel) %>%
  ebPredict(data_formatted) %>%
  ebSummary %>%
  ebSavings(deemed = ebSample_deemed)

The results show that the measures achieved savings of 2,374 kWh and 1,550 at the two sites, signifying realization rates of 0.79 and 1.55 when compared to deemed savings. Energy savings and loadpaths can be visualized using the ebPlot() function.

Lastly, the intermediate ebSummary() function can be used to evaluate metrics that indicate the quality of the models at the meter and site level (such as CVRMSE):

ebList(data_formatted)[['baseline']] %>%
  lapply(ebModel) %>%
  ebPredict(data_formatted) %>%
  ebSummary 

aroyal641/ebase documentation built on Oct. 3, 2020, 4:03 a.m.