1.Introduction

Forest City residents in Hawaii are experiencing a hybrid billing strategy where they pay for the extra electricity they use. “Extra” electricity usage occurs when a resident energy consumption is over the 70th percentile for their housing group, as shown on the Figure below. Oppositely, they receive a rebate if they are below the 30th percentile. This has led to an interest in achieving more energy savings by the residents.

Figure 1.1 Location of the 15 houses in the study. The houses were slight repositioned for privacy of the residents.

Figure 1.1 Location of the 15 houses in the study. The houses were slight repositioned for privacy of the residents.

1.1 What’s New?

This report reviews and extend analysis of the previous 3 reports as follows:

  • New: Survey questions, sensor data rationale and relationship between both has been added.
  • New and Revised: Limitations of the study across the 3 reports have been organized in a single section (3.2). Additionally, assumptions omited in the previous 3 reports have been made explicit in this report as threats to validity (sections 3.2.2 , 3.2.3, and 3.2.4) when handling missing, inconsistent or data transformations for analysis (section 4).
  • New: Time Series Clustering Analysis for Appliance’s Manufacturer Comparison.
  • New: Thermal Comfort Analysis of various rooms using ASHRAE55 and Berkley Models.
  • New: Data Model to simplify the question What data is available? (Section 3.1) enabling usage for other analysis.
  • New: Minimal explanation of the purpose of each sensor has been added, so non-domain experts can also use the data.

1.2 Report Outline

The remainder of this report is divided as follows:

2. Method

2.1 Sensors

2.1.1 Solar Water Heating System

In order to achieve more energy savings, the 15 house residents have been given the option to heat water with sunlight instead of a electric heater. The main purpose of the solar water heating system is to provide an adequate water temperature for the least amount of energy expenditure by the backup electric heater.

The solar water heating system is shown on Figure 2.1.

Figure 2.1 Solar Water Heating System.

Figure 2.1 Solar Water Heating System.

The system in Figure 2.1 function as follows: Water is stored in a tank with 3 pipes (a,b, and c). Water flows out (b) of the tank to the roof, where the solar panels are located, the water is heated, and flow back (a) to the tank. The tank’s water is used by the resident when flowing through the upper pipe (c). The flow of water to the roof is controled by the solar recirculation pump. When the pump is on, we know the resident opted to use the solar water heating system to heat the house’s water. Alternatively, we can observe the energy consumption associated to the electric heater to know when it is being used.

2.1.2 Water Usage for Laundry and Meals

To better understand residents decision between using the solar water heating system or the electric heater, a follow-up step is understand the residents habits in using the water for various activities in the house. This has been done by monitoring usage of the following appliances in each household:

  • 1.1 Clothes Washer
  • 2.1 Dish Washer

Possible associated to those two devices usage are also:

  • 1.2 Dryer
  • 2.2 Stove

When a resident is doing laundry or preparing/finishing a meal.

2.1.3 Water Usage for Thermal Comfort

Hot water may be used for a shower, which may be conditioned to the resident’s thermal comfort level. In order to assess the thermal conditions and expected comfort, temperature and relative humidity were monitored in diferent locations inside the house, and outside the house.

To observe how the resident would react to the different thermal conditions, the following appliances usage were also monitored:

  • AC
  • Fan

2.1.4 Appliance’s Manufacturer Comparison

The appliances(e.g. AC/Dryer/Clothes Washer) parts manufacturers were also logged for comparing different energy costs to achieve the same functionality (e.g. Two different ACs having different energy costs for the same usage, in which case savings could be obtained by replacing the higher energy consumption AC).

Table 2.1 Air conditioning system components.

Table 2.1 Air conditioning system components.

Table 2.2 Other major appliances.

Table 2.2 Other major appliances.

2.2 Survey

Surveys were applied once to each house to compare residents frequency and duration of appliance usage, satisfaction with the solar heating system, thermal comfort preferences and management of comfort (using AC and/or Windows), and demographics of each household for house energy consumption comparison. This section describes how each set of questions in the survey complement data collected from the sensors.

2.2.1 Solar Water Heating System

To complement resident’s usage understanding between the solar water heating system and the electric heater, we surveyed participants for their satisfaction (Q.18.2).

  • 1.1 Solar Water Heating System
    • (Q.18.2) Do you get enough hot water?

2.2.2 Water Usage for Laundry and Meals

For appliance water usage, we surved residents on their usage assumptions of dishwasher (Q.10) and clotheswasher (Q.16.1) frequency to compare against the actual usage reported by sensors.

We compare dishwasher usage duration by verifying the load quantity (Q.11). To complement understanding of when water is still used for the same purpose and not measured by sensors, we also asked for hand wash dishes frequency (Q.12).

We compare laundry usage duration by again verifying the load quantity (Q.16.1). To complement understanding of water heating usage by residents, we surveyed about residents frequency of using cold (Q.16.2), warm (Q.16.3) or hot water (Q.16.4), not explicitly captured by sensors.

  • 1.2 Water Usage for Laundry and Meals
    • (Q.10) Dishwasher: how many loads/week?
    • (Q.11) Dishwasher: are they full or partial loads?
    • (Q.12) How many times/week do you hand wash dishes?
    • (Q.16.1) Number of loads of laundry per week?
      • (Q.16.2) # cold
      • (Q.16.3) # warm
      • (Q.16.4) # hot

As noted in the related section to Water usage for laundry and meals (2.2), we also extended the water usage monitoring to appliances that are used paired to dishwashers and clothes washer: stove (laundry) and dryers (meals). Therefore, the survey also complements those two appliances.

We surveyed usage assumptions the dryer (Q.17.1) and to compare duration, the dryer’s load size (Q.17.2, 17.3), and stove (Q.13,14,15).

Similar to complementing the usage of dishwasher by observing how often hand dishwashing was done, here we also surveyed how often the microwave is used (Q.15). Since fridges and freezers also consume a high amount of energy, and relate to meals, we also surveyed any extra number of them in the house (Q. 9).

Observe that the way duration questions were asked differ between explicitly asking how long it is used, or implicitly by asking the load quantity, as it is more likely a person is able to estimate explicitly the duration of meals to get prepared instead of being inferred by quantity, while the opposite is true for laundry.

  • (Q.17.1) Loads per week in clothes dryer?
    • (Q.17.2) # small
    • (Q.17.3) # large
  • (Q.13) Number of mins/day using stove’s burner? (The question says range on the question file and burner on the answer file, which is it?)
  • (Q.14) Number of hour/week using stove’s oven? (Original question was How many hours/week do you use your oven?)
  • (Q.15) How many min/day do you use the microwave?
  • (Q.9) Do you have a second frig or freezer?

2.2.3 Water Usage for Thermal Comfort

For thermal comfort asessment, we surveyed the residents thermal comfort temperature preference through their thermostat settings (Q.3, Q.4) and management of thermal comfort when uncomfortable through usage of AC and windows (Q. 5, Q. 6, Q.7).

  • 1.3 Water usage for Thermal Comfort
    • (Q.3) Thermostat setting
    • (Q.4) If you set temperature higher at night or while out, what temperature?
    • (Q.6) Do you open the windows for fresh air?
    • (Q.7) If open windows, do you turn off AC?
    • (Q.5) Is the AC on all of the time, some of the time…

Additionally, showering can also be used to manage thermal discomfort, and was also surveyed (Q.8.1, 8.2, 8.3).

  • (Q.8.1) How many showers per week?
  • (Q.8.2) Average length of each shower?
  • (Q.8.3) How many baths per week?

2.2.4 Appliance’s Manufacturer Comparison

To complement the appliance’s manufacturer comparison, impressions from residents on malfunctioning devices were also surveyed to help triangulate energy profile outliers from the associated appliances.

  • 1.4 Manufacturer’s Device Energy Profile

    • (Q.18.1) Does the AC work okay?
    • (Q.18.3) Does the dryer work okay?

Finally, demographic data was also logged to provide a better baseline on energy consumption and observed savings when comparing houses in and between neighborhoods:

  • (Q.2.1) Total Number of Residents
  • (Q.2.2) Number of 0-10y Residents
  • (Q.2.3) Number of 11-18y Residents
  • (Q.2.4) Number of Adult Residents
  • (Q.19) Number of months in house

To-Do: Clarify the following question meaning/purpose:

  • (Q.1) Number of stores of home? (All answers were “2”)

3. Dataset

3.1 Data Model

The collected data model is shown on Figure 3.1.

Figure 3.1 Data Model.

Figure 3.1 Data Model.

Every house contains a few rooms, appliances, and a water tank . Houses were also surveyed once, had visits for various reasons, and had their house coordinates approximated to preserve the participants privacy.

The water tank is part of the solar water heating system, as discussed in section 2.1.1, and it’s 3 pipes water temperature and pump being on/off were measured. For houses 1, and 2, the water flow was measured instead of the pump usage. Unfortunately, measurement validation yielded water flow inacurracy of over 44.2%, and the data was discarded in this data model. House 6 never used an electric heater, and can be used to see how the water temperature correlate to energy costs. The data in this table is provided by the Apollo sensors.

In different rooms of every house were placed sensors to measure temperature and relative humidity, in order to assess thermal comfort. One of the sensors was also placed outside the house, as discussed in section 2.1.3. The data in this table is provided by the HOBO sensors.

In a given house, different appliances (uniquely identified by type and manufacturer) energy consumption were monitored, and it’s parts manufacturer logged in a data_sheet, as shown on Tables 2.1 and 2.2, allowing for usage monitoring (e.g. AC, dishwasher, etc). The number of appliances monitored per house was bounded by the 12 available current transducers (cts). Houses 1,4,7 and 8 had one of their cts used on the tank’s pump for sensor’s reading trinagulation, as shown in Table 3.1. The data in the appliances table is provided by the eGauge2 sensors.

Table 3.1 The 12 Current Transducers (cts) placement in each house.

Table 3.1 The 12 Current Transducers (cts) placement in each house.

From Table 3.1, we can observe not all appliances energy monitoring will be available in these 4 houses. Namely, houses 1,4, and 7 do not contain dish washer energy consumption readings, while house 8 does not contain clothes washing energy consumption readings.

3.2 Threats to Validity

3.2.1 House Infrastructure

Table 3.1 Start and end dates for energy data collection for this study.

Table 3.1 Start and end dates for energy data collection for this study.

3.2.2 Apollo Data Collection

3.2.3 eGauge2 Data Collection

3.2.4 HOBO Data Collection

4. Analysis

4.? CBE Thermal Comfort Analysis

house.id <- 16 #Only house 16 has hobo data......which is not even included among the 15 houses on the reports. How come?
hobo <- fread("~/Desktop/fcphase3/environment.csv",header=TRUE)
house.comfort <- hobo[house_id == house.id]
a <- house.comfort[,.(temp_f,temp_f,rh_pct)]
colnames(a) <- c("ta","tr","rh")
a$met <- 1.1
a$clo <- 0.5
a$wme <- 0
a$vel <- 0.1


#Load Google's Javscript Engine V8 (See https://cran.r-project.org/web/packages/V8/vignettes/v8_intro.html) 
library(V8)
#Create a new context
ct <- v8()

#Load Javascript Library for forEach function
ct$source(system.file("js/underscore.js", package="V8"))
#Load local comfortModel javscript library (only modified the path of the libraries)
ct$source("comfortmodels2.js")
ct$source("util.js")
ct$source("psychrometrics.js")

#PMV ranges from -3 to 3 according to ASHRAE55 2013 being respectively:
#“cold,” “cool,” “slightly cool,” “neutral,” “slightly warm,” “warm,” and “hot.”

#Apply the function over all the table
#pmv <- data.table(ct$call("_.map", a, JS("function(x){return(comf.pmvElevatedAirspeed(util.FtoC(x.ta),util.FtoC(x.tr),x.vel,x.rh,x.met,x.clo,x.wme))}")))

#Have to be careful with the order of parameters in comfortmodels.js and the units (defined on the documentation). 
pmv<- data.table(ct$call("_.map", a, JS("function(x){return(comf.pmv(util.FtoC(x.ta),util.FtoC(x.tr),x.vel,x.rh,x.met,x.clo,x.wme))}")))

#See https://www.r-bloggers.com/new-in-v8-calling-r-from-javascript-from-r-from-javascript/ to use the console to inspect individual functions. 

house.comfort$pmv <- pmv$pmv

hist(house.comfort[room_id == 1]$pmv) #Living Room
hist(house.comfort[room_id == 2]$pmv) #Master Bedroom
hist(house.comfort[room_id == 3]$pmv) #Bedroom 2
hist(house.comfort[room_id == 4]$pmv) #Attic
hist(house.comfort[room_id == 5]$pmv) #Outside
hist(house.comfort[room_id == 6]$pmv) #Kitchen
hist(house.comfort[room_id == 7]$pmv) #Garage
hist(house.comfort[room_id == 8]$pmv) #Master bath

#west.pmv <- data.table(ct$call("_.map", test2, JS("function(x){return(comf.pmvElevatedAirspeedutil.FtoC(x.ta),util.FtoC(x.tr),x.vel,x.rh,x.met,x.clo,x.wme))}")))

5. Conclusions

The report conclusions were that further inspection on the data collection routines were necessary before moving forward with any analysis. Nonetheless, the reorganization of the method presented in this report along with associated rationale can be useful in replication to further studies in this theme, and definition of the data model of the database, as it was done here.