GREEN Grid Choice Modelling and Questionnaire: Results

John Williams

August 26, 2014

Overview of progress

Number of responses:

  • PV: n ≈ 1018 of 1,000
  • HEMS: n ≈ 515 of 500
  • EV: n ≈ 505 of 500

Note: I've written ≈ above becuase I've not yet checked for valid responses

Agenda

  • Sample profile
  • HEMS
  • PV
  • EV

For each of the three technologies, we'll look at:

  • Implicit or “revealed” preferences (based on utilities from choice modelling)
  • Stated preferences (responses to questionnaire)

Sample profile: age and sex

Population comparisons from the 2013 Census

Age group Sample % Pop. %     Δ
18-24 11 10 1
25-34 8 13 -5
35-44 16 13 3
45-54 22 14 8
55-64 21 11 10
≤ 65 22 14 8
Sex Sample % Pop. %   Δ
Male 47 49 2

Sample profile: personal income

Group Sample % Pop %   Δ
Loss 0.3 0.6 -0.3
Zero income 2.8 7.9 -5.1
$1 - $5,000 2.6 6.1 -3.5
$5,001 - $10,000 2.7 5.4 -2.7
$10,001 - $15,000 4.9 8.8 -3.9
$15,001 - $20,000 3.9 9.4 -5.5
$20,001 - $25,000 6.2 7.4 -1.2
$25,001 - $30,000 5.4 6.3 -0.9
$30,001 - $35,000 3.9 5.7 -1.8
$35,001 - $40,000 4.7 6.2 -1.5
$40,001 - $50,000 10.0 9.5 0.5
$50,001 - $60,000 9.4 7.4 2.0
$60,001 - $70,000 9.6 5.5 4.1
$70,001 - $100,000 16.6 7.8 8.8
$100,000 - $150,000 10.3 3.6 6.7
$150,001 or more 6.8 2.3 4.5
Declined 12.8 9.7 3.1

Sample profile: education

The sample is not representative by education:

  • ≈20% of NZers (aged 15 and over) have post-secondary education
  • 72% of the sample have post-secondary education

Sample profile: region

Region Sample % Pop %     Δ
Northland 2 4 -2
Auckland 22 34 -12
Waikato 6 9 -3
Bay of Plenty 5 6 -1
Gisborne & Hawke's Bay 3 4 -1
Taranaki 2 2 0
Manawatu-Wanganui 4 5 -1
Wellington 7 11 -4
Nelson/Marlborough/Tasman 5 3 2
West Coast 1 1 0
Canterbury 31 13 18
Otago 11 5 6
Southland 2 2 0

Sample profile: summary

The sample is representative by sex, but nothing else:

  • Biased toward older, higher income and highly educated people
  • Heavily biased toward Cantabrians, and severely under-representative of Auckland

Implications:

  • Age and income bias is not necessarily a bad thing, as the target market for HEMS, PV and EVs is higher net worth/disposable cashflow people
  • Geographical bias is probably problematic

Knowledge of technologies

Knowledge level PV EV HEMS
I don't even know what those words mean, really 30 1 6
I know what this is, but that's about it 15 10 18
I know a little, but probably less than most people 10 19 23
I know more than most people about this 14 11 9
I probably know as much as most other people about this 31 59 44

Summary: lack of knowledge about PV (very surprising!)

Interest in technologies

Knowledge level PV EV HEMS
I've never considered purchase 39.2 38.1 44.7
I 've thought about it but rejected the idea 19.2 27.2 13.5
I'm still thinking about it 28.5 28.1 32.2
I'm almost ready to buy 5.3 3.5 3.8
I'm ready to buy 3.5 2 2.5
I already have one 2.7 0.5 2.7
I had one in the past, but not now 0.7 0.2 0.3
I've got one, and am thinking of buying another 0.8 0.3 0.2

Note: 32 respondents have a PV system and 72 are ready to buy, and about half in each category are willing to be interviewed Summary: few differences across technologies, possible exception of rejecting buying an EV

Willingness to be interviewed

Technology No Yes % Yes
HEMS 306 200 40
PV 622 346 36
EV 335 158 32

A surprising proportion of respondents indicated a willingness to participate in a follow-up interview

Choice Modelling: interpreting results

  • The output of CM is the utility of each level of each attribute for each respondent
    • Utilities have no absolute meaning; they can only be interpreted relative to each other (hence the utility of one level is set to 0 for each attribute)
    • The sum of the utilities of all the levels of each attribute can be interpreted as the utility of the attribute
    • The sum of each utility across all attributes is the utility of the alternative
  • Primary use of utilities is …
    • in academia, to estimate the value of each attribute
    • in commercial use, to estimate the value of each alternative

HEMS choice modelling

  • Time of day of interruption
    • Any time
    • Agreed with service provider
    • Under total control of customer
  • Length of interruption
    • ≤ two hours
    • ≈ 30 minutes
    • ≤ 5 minutes
  • Notice of interruption
    • ≤ 1 minute
    • ≥ 30 minutes
    • ≤ two hours
  • Financial return
    • None
    • $50 pa
    • $200 pa
  • Control over interruption
    • Automatic, no override
    • Automatic, with override
    • Manual

Utilities of HEMS levels

plot of chunk unnamed-chunk-8

(Time, length, notice, return, control)

Utilities of HEMS attributes

plot of chunk unnamed-chunk-9

Differences significant on pairwise t and Wilcoxon tests

Appetite for HEMS

If putting some of your appliances under automatic control would save you $100 per year, would you consider it?

Response %
Yes 47
Maybe 44
No 9

Appliances under HEMS control?

Appliance Never Manual control Auto with override Auto no override Already controlled
Clothes dryer 7 24 45 13 11
Dishwasher 7 28 44 12 8
Clothes washer 10 30 44 7 9
Under floor 9 24 39 10 18
Heat pump 24 25 38 1 11
Water heater 9 39 31 11 9
Towel rail 5 23 27 24 21
Fridge 46 25 22 5 3
Freezer 44 26 20 6 3
Fridge/freezer 47 25 20 5 3

Communication preferences

“If you were to receive messages from your electricity provider asking you to turn certain appliances off or notifying you that they are going to turn an appliance off for a specific period of time (e.g. in return for some kind of reward) how would you prefer to receive these notifications?

Please tick all that apply”

40% of respondents chose more than one option.

Medium Endorsed (%)
SMS 52
Email 35
App 38
Device 42

Control preferences

Controlling party Peferred (%)
Energy company 39
Another company 1
Customer 60

Summary: HEMS

  • Surprisingly, financial benefit seems to have least utility, perhaps because it is so low
  • Although majority prefer customer control, substantial minority willing to cede control
  • Almost half say “yes” to HEMS, a bit less “maybe”
    • only 9% say “no” to HEMS
  • Substantial minority willing to cede control of dish and clothes washers & dryer
  • Substantial minority would never cede control of fridge & freezer

PV choice modelling

  • Upfront cost of system
    • $4,000
    • $8,000
    • $16,000
  • Payback period
    • 5 years
    • 10 years
    • 15 years
  • Grid dependence
    • Disconnect a few days p.a.
    • Disconnect most days
    • Completely disconnect
  • Aesthetics of panels
    • Big and highly visible
    • Small and discreet
  • Ownership of system
    • Owned by electricity company
    • Leased from electricity company
    • Owned by customer

Utilities of PV levels

plot of chunk unnamed-chunk-15

(Upfront, payback, aesthetics, dependence, ownership)

Utilities of PV attributes

plot of chunk unnamed-chunk-16

Differences significant on pairwise Wilcoxon tests

PV barriers & benefits

Stated importance of barriers and motivators was measured by these questions:

  • Imagine that, for some reason, you have definitely decided to install solar panels on your roof. What would be the biggest problem that you would face?

Allocate points to each reason, where more points indicate more importance to you. You can either drag the sliders or type in the boxes. Make sure that the allocated points add to 100 (you can't proceed unless they do).

  1. The financial cost
  2. The effort involved with choosing the right system for me and my dwelling
  3. The difficulty of accurately estimating the financial returns
  4. Safety and security
  • What, to you, would be the biggest benefit of solar PV installed at your dwelling?

Allocate points to each reason, where more points indicate more importance to you. You can either drag the sliders or type in the boxes. Make sure that the allocated points add to 100 (you can't proceed unless they do).

  1. Independence from the national grid
  2. Reduced electricity bills
  3. Contribution to New Zealand's sustainable energy supply

PV barriers & benefits

plot of chunk unnamed-chunk-17

(Allocate points to each barrier or incentive such that they sum to 100)

All different on pairwise Wilcoxon tests

Summary of PV

  • Biggest barrier is upfront cost (duh!), however importance of this attribute is most heterogeneous
  • Biggest incentive is reduced bills; grid independence and sustainability about equal
    • Importance of upfront cost and reduced bills about equal
  • Aesthetics have lowest utility, then ownership; other attributes about equal

EV choice modelling

  • Purchase price
    • Twice as much as non-EV
    • 50% more than non-EV
    • Same price as non-EV
  • Ongoing costs
    • 20% less than non-EV
    • 30% less than non-EV
    • Half as much as non-EV
  • Range
    • Half the range of non-EV
    • 75% the range of non-EV
    • Same range as non-EV
  • Age of vehicle
    • More than 5 years old
    • Less than 5 years old
    • New
  • Charge time
    • 6 hours
    • 30 minutes
    • 5 minutes

Utilities of EV levels

plot of chunk unnamed-chunk-18

(Price, ongoing, range, age, charge)

Utilities of EV attributes

plot of chunk unnamed-chunk-19

Differences significant with pairwise t and Wilcoxon tests

EV purchase likelihood (money no object)

“Imagine that you have been given $100,000 but you could only spend it on a new car (and could not keep the change!). How likely is it that you'd buy an electric vehicle?”

Response %
I'd never buy an EV 8
Maybe 31
Would seriously consider 35
Likely 16
Definitely 10

EV incentives to buy: fuel cost

Response Current 50% more Twice as much
Never 20 7 5
Maybe 45 24 12
Consider 22 38 26
Likely 9 21 32
Definitely 4 10 26

EV incentives to buy: vehicle cost

Response Twice as much The same Half as much
Never 62 6 2
Maybe 30 23 9
Consider 6 40 21
Likely 2 22 27
Definitely 1 8 42

EV incentives to buy: range

Response Half the range 75% of the range The same range
Never 37 8 3
Maybe 43 32 15
Consider 15 39 29
Likely 4 16 31
Definitely 0 4 23

Summary of EV results

  • Purchase price is most importance attribute
  • Range and charge time second-equal
  • Only around 10% of respondents would “definitely” buy an EV, even if it was essentially free
    • however about 61% would seriously consider, be likely to buy or definitely buy
    • and only 8% would never consider buying
  • Fuel cost seems to have a large influence on purchase likelihood, about the same as upfront cost

Conclusion

  • Analysis has only just begun: descriptive results only have been produced so far
    • Will investigate differences, associations and clustering by utilities later this year
    • Suggestions for further analysis are welcome
  • These slides are available at http://rpubs.com/johnfrombluff/GGCMPR
    • Will be updated as new results become available
  • You can email me at john.williams@otago.ac.nz if you have any questions, suggestions or comments