OMF Calibration

= Background = After populating the loads onto the feeder, the size and behavior of the load must be adjusted to match available circuit level data. The overall idea behind calibration to SCADA data was to create an automated script which populated the feeder model with building and load models that formed a random distribution of pre-defined characteristics, such as building insulation, load magnitude, etc. However, where appropriate, these values were attached to a parametric adjustment value that allows the user to calibrate to the time-series annual demand. These parametric values included such things as the relative number of buildings modeled, percentage penetration of HVAC and heat sources, relative magnitude of commercial loads versus residential loads, ZIP fractions, relative magnitude of hot water demand, and more. Most of these values are collected from available data, while allowing a slight adjustment to more accurately match the SCADA data.

= Previous Work = The original work was performed manually by adjusting lumped parametric values to adjust detailed distributions of parameters. This required "engineer-in-the-loop" to understand what adjustments were needed and could be very time-consuming.

Daily, weekly, and seasonal schedules were created to control thermostat set points within the homes. These were created to loosely represent a variety of customers, including those who leave their settings the same throughout a season, those who adjust the set points only on weekends, and those who adjust them on a daily or hourly basis (away versus awake versus asleep). These schedules were created from a combination of survey data and randomly distributed throughout the population of residential buildings. Adjustments were made to represent differences between seasons, between daytime and nighttime, and between weekends and weekdays, and each building contained its own unique schedule. Commercial buildings were assumed to keep more constant thermostat settings, with adjustments only made between their daytime and nighttime settings, and used similar settings for both weekdays and weekends. While the commercial settings were more constant, there were still variations between weekday and weekend to represent the behavior of commercial office buildings.

Hot water demand schedules were created to represent the amount of demand by hot water heaters. These were created from a combination of survey data, Department of Energy (DOE) water heater loading approximations, and ELCAP load shapes. Events related to showers, dishwashers, hand washing, and clothes washers were simulated to represent the demand on the hot water heater. Once again, each building with an electric hot water heater (buildings supplied by gas lines were assumed to have gas water heaters) contained its own unique water demand schedule. The magnitude of the hot water heater demand, and hot water heater penetration level, were configurable parameters that could be adjusted to increase or decrease the water heater usage. It should be noted that this could be replaced with other data, such as that available from DOE for standard hot water usage.

These two loads were selected as state models (as opposed to the more generic ZIP models) because of their large impact on the demand of a residential home. Capturing the actual state driven behavior, as opposed to average behavior, was essential in understanding the effects of voltage reduction because the average behavior has not been fully quantified during voltage reduction operations. To capture the effects of smaller appliances, time-varying ZIP models were created. The time dependent effect was created using a library of yearly load shapes, containing seasonal, weekly, daily, and hourly variations at 15-minute intervals, most created from raw SCADA data.

While in the first iteration, appliances were represented in a lumped model, in subsequent development, these load shapes were broken into more detail, including load shapes for several appliances which could each be individually adjusted in magnitude and time. The need of the study should dictate how much detail is given at the appliance level. For example, when studying the effects of voltage, a lumped, reduced-order model was able to capture the effects; however, if the study is looking at the effects of the load shifting behavior of a demand response enable clothes dryer, then individual models are more appropriate. Large commercial and industrial loads were created using a similar method to the smaller appliances, using available SCADA information. Power factor and ZIP fractions were assigned from available anecdotal information, including measured laboratory data and previous CVR studies.

= Current OMF Work =

= See also = OMF_Scripting_Documentation

OMF Conversion Process

OMF Population Process

OMF Weather Extractor