The TOU project - Test 1

Adaptive and TOU Pricing Schemes
for Smart Technology Integration
ORDECSYS
Christopher Andrey
2014
An overview of the results

ORDECSYS
»The TOU Project - An overview
Aim of the project:
Assess the influence of smart grid technologies
(decentralised storage and demand-response) on the
long-term planning of a regional energy system
Funded by: Consortium:
(Forschungsprogramm Energie - Wirtschaft - Gesellschaft)
ORDECSYYS

{ Higher energy efficiency
More renewables
2050** Factor
PV 320 GWh 11.4 TWh x35
Wind 88 GWh 4 TWh x45
ORDECSYS
+
2012
Production*
Swiss Target
* Statistique suisse de l’électricité 2013, SFOE ** Presentation by Pascal Previdoli, SFOE
+
Nuclear Phase-Out GHG Emission Reduction
In particular, the Swiss Energy
Strategy 2050 massively relies on
investments in intermittent
renewables

ORDECSYS
One of the bottlenecks for a wide-spread
penetration of renewables is their
intermittent production pattern.
Solar
Wind
Source : http://www.transparency.eex.com/

Leitstudie 2009 national ohne zusätzliche Verbraucher –
ORDECSYS
Seite 8
Leitstudie 2009 national ohne zusätzliche Verbraucher –
2050 (meteorologisches Basisjahr 2007)
(meteorologisches Basisjahr 2007)
Demand-Response
German load curve in 2050
Dr. Kurt Rohrig, Fraunhofer-Institut, Kassel
PV
Hydro
Biomass
Geothermal
Wind
Others
Storage

ORDECSYS
Both storage and demand-response may be achieved through
time-dependent financial incentives.
Load reduction vs LMP in PJM (USA)
greentechmedia.com (peak ~ 160 GW)

!
!
ORDECSYS
Measure of the attractiveness of demand-response
and decentralised storage in electric vehicles 10
Scenario 1 Scenario 2 Scenario 3
Appliance Dishwasher Dryer Fr e e z e r
Control Method Own Computer Manual Network Operator
Delay 6 hours 10 min 2 hours
Yearly Incentive 50 CHF 10 CHF 0 CHF
! Choice X
Table 2: Demand-Response Evaluation - Example {Table:DR-Ex}
2.5 Storage in Electric Vehicles
1.0
The aim of the third and final part of the survey is to understand under which cir-cumstances
respondents would agree to put their electric car at the disp osal of the
0.8
batteries as temporary storage
use the cars’ that the latter 0.6
could incentives, of
network operator so estimating the role of financial interested in to
are units. In particular, we guaranteed autonomy after participating the electric
ownership model of the battery, 0.4
of to be connected to the the duration the car has the minimum such a service, and of network per day.
The respondents have again been introduced to the subject via a short home-made
factual animation, embedded in the survey environment by LINK. Clicking on Figure
2 will open the animation on YouTube. In this case to o, the resp ondents were asked
(i) to imagine themselves living in 2030 and (ii) to imagine owning an electric car.
scenario s) =
prob(exp
P
l2levels(s)
wl
P
t2scenarios
exp
P
k2levels(t)
wk
-5 0 5 10
through conjoint analysis techniques
0.2
16
Quite surprisingly, the amount of time by which the consumption is shifted has
almost no influence on the probability of a given scenario.
3.2.4 Yearly Incentive
Figure 8: Part-worths of the yearly incentive attribute. Source: Annex D. {DR-U4}
Finally, the impact of yearly incentive on the choices reveals that o↵ering only a
modest amount of 20 CHF per year seems to dramatically increases the probability of
adoption.
The methodology adopted in the second part of the survey has allowed us to measure
the part-worths of each of the levels of the attributes. Let us consider each of the
attributes in turn:

ORDECSYS
Sample:
A total of 1045 respondents from
‣ Canton of Geneva (373)
‣ Canton of Vaud (367)
‣ Cantons of Neuchâtel, Fribourg, Jura (305)
Ages ranging from 15 to 74
Online survey (internet users)
Survey rolled out between Nov. 4th and Nov. 18th, 2013
Introduction to the ES2050, DR and storage in two short animations

Simulations obtained using a randomised first choice model. Acceptance here means the probability of adopting the given scenario rather than choosing “None”.
ORDECSYS
Conjoint Simulations
Lave-vaisselle 1/2
80.1
%
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Acceptance
Appareil Lave-vaisselle
Contrôle Par votre
Amplitude de
déplacement 10 minutes
Impact sur la
facture annuelle CHF 0
N = 1048 respondence
ordinateur
13 | 10.02.2014 | TOU Pricing Ordecsys
100%
90%
80%
70%
60%
50%
73.7% 80.1% 76.7%
100%
80%
60%
40%
20%
0%
Contrôle
40%
30%
20%
10%
Manuel Par votre ordinateur Par votre distributeur
Appareil Lave-vaisselle
d’électricité
Contrôle Par votre
ordinateur
Amplitude de
déplacement 10 minutes
Impact sur la
facture annuelle CHF 0
80.1% 79.6% 79.3% 78.5% 81.2%
100%
80%
60%
40%
20%
0%
Amplitude de déplacement
10 minutes 30 minutes 1 heure 2 heures 6 heures
21
Lave-vaisselle 2/2
80.1% 84.2% 85.1% 86.8%
100%
80%
60%
40%
20%
0%
Impact sur la facture annuelle
CHF 0 CHF 10 CHF 20 CHF 50
80.1
%
0%
Acceptance
Figure 14: Utilities of the yearly incentive attribute. Source: Annex E. {DR-S2}
Another interesting simulation is the one related to the flexibility of dryers’ use. The
scenario Dryer, Own Computer, 10 minutes, CHF 0 has a rather low acceptance of
72.7% as shown on the LHS of Figure 15. Remember that the dryer had the lowest
utility, see Figure 5. However, o↵ering 50 CHF per year can increase the acceptance
by almost 10 percentage points, as can be noticed on the RHS of Figure 15.
Sèche-linge 2/2
72.7
%
100%
90%
80%
Results:
‣ ~80% acceptance
‣ low sensitivity to the
implementation details

ORDECSYS
22
Propriété du ménage 1/2
Autonomie garantie 400 kilomètres
Durée de la mise à
disposition par jour 1 heure
Gains annuels 100
83.8
%
100%
Generally speaking, the acceptance of temporary storage in electric cars is very well
accepted as can be seen on the simulations presented in Figure 16 and 17.
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Propriété de la
batterie
83.8
%
Gains annuels 100
Propriété du ménage 2/2
Propriété du
ménage
100%
80%
60%
40%
20%
100%
80%
60%
40%
20%
40%
20%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Acceptance
100%
80%
60%
40%
20%
Simulations obtained using a randomised first choice model. Acceptance here means the probability of adopting the given scenario rather than choosing “None”.
Acceptance
80.3% 80.9% 83.2% 83.8%
0%
Autonomie garantie
100 kilomètres 150 kilomètres 250 kilomètres 400 kilomètres
83.8% 83.5% 83.7% 82.4%
0%
Mise à disposition par jour
1 heure 2 heures 6 heures 12 heures
Figure 16: Utilities of the yearly incentive attribute. Source: Annex E. {EV-S1}
83.8
%
100%
90%
80%
0%
1 heure 2 heures 6 heures 12 heures
Figure 16: Utilities of the yearly incentive attribute. Source: Annex E. {EV-Propriété du ménage 2/2
83.8% 86.4% 87.8%
0%
Gains annuels
Propriété de la CHF 100 CHF 300 CHF 700
batterie
Propriété du
ménage
Gains annuels 100
Figure 17: Utilities of the yearly incentive attribute. Source: Annex E. {EV-All combinations of levels give rise to acceptabilities that are in the 80% range.
Rue du Gothard 5 – Chˆene-Bourg – Switzerland Tel. +41 22 940 30 20 – www.ordecsys.com
Results:
‣ ~84% acceptance
‣ low sensitivity to the
implementation details

ORDECSYS
ETEM-SG is a long-term energy planning (LTEP) model:
‣ used to assess the impact of regional energy/climate policies
‣ represents the entire energy system of a region
‣ embeds a detailed representation of
‣ technologies (investment costs, O&M costs, efficiency, etc.)
‣ demands for energy services in all sectors (residential, industry, etc.)
‣ dynamics of the demands

ORDECSYS
Comparaison simulation - courbes de charges réelles
Résidentiel collectif - Printemps
80
4000
Period 1 Period i Period N
typically 1 to 5 years
Calibration Investment i
70
60
50
40
30
20
10
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Horizon: typically 10 to 50 years
3500
3000
2500
2000
1500
1000
500
0
Autre électroménag
Clim
Chauff appoint
Chauff principa
Veilles
Plaques de cuisson
Fours micro-ondes
Fours traditionnel
E.C.S.
Informatique
Congél
Combinés
Réfrig
Eclairage
Sèche-linge
Lave-vaisselle
Lave-linge
Téléviseur
Total réel coll
568
LES BOUDINES;28/05/2002
862
AVANCHET PARC RTE DE MEYRIN; 12/07/2002
675
LIGNON EST; printemps 2002
Détente-SIG-c_zone_hab.xls
demand allocation
(dynamics)
General time-structure

EV Imported
biomass
ORDECSYS
General structure
CHP
Electricity
Heat
CO2
Transport
demand
Large-scale flow problem

minimise total cost
flow conservation
technology description
activity bounded by capacity
other constraints (e.g. CO2)
ORDECSYS
Annexes
Modèle déterministe
General mathematical structure
X = flows, C = capacity increase, I = imports, E = exports
i,j = commodity index, t = time index, k = technology index
Soit i et j les indices des commodités, k l’indice des technologies et t l’indice des périodes, la
formulation mathématique simplifiée du modèle ETEM s’écrit:
min f(X,C, I,E) (1a)
Iit +
X
k
Xout
ikt = Eit +
X
k
Xin
ikt + dit, 8i 8t (1b)
X
j
!ijktXin
jkt = Xout
ikt , 8i 8k 8t (1c)
X
i
Xout
ikt  ↵kt#kt(ckt +
X
lt
Ckl), 8k 8t (1d)
gm(X,C, I,E)  0, 8m (1e)
avec X = (Xin,Xout), les variables représentant les flots de commodités entrant et sortant
des technologies, C les variables d’investissement dans les capacités de technologies et I et E
les variables d’import et d’export. La fonction objectif f(X,C, I,E) représente l’ensemble des
coûts et profits annualisés fixes et variables associés aux technologies et à leur utilisation, aux

ORDECSYS
{ Current energy
system
(capacities)
Evolution of useful demands
and of imported energy prices
(drivers)
Catalogue of existing
and future technologies
ETEMSmartGrid
Sources of uncertainties
‣ Capacity expansion (technology portfolio)
‣ Activities (operation)
‣ GHG and pollutants emissions
‣ Imports and exports
‣ Marginal costs (electricity, GHG, etc.)
1
2
3
4

ORDECSYS
Cantons of Vaud & Geneva
2005-2050
Inputs
1. Current energy system
Hydro VD
Hydro GE
PV
Cheneviers
Tridel
Pierre de Plan
Chatillon
Veytaux
0.9"
0.8"
0.7"
0.6"
0.5"
0.4"
0.3"
0.2"
0.1"
0"
""""""WN"" """"""WP1""""""""WM""""""""WP2"" """"""SN"" """"""SP1"" """"""SM"" """"""SP2"" """"""IN"" """"""IP1"" """"""IM"" """"""IP2""
Electricity production 2005 Load curve 2005
Transport
Heat & Warm Water
Industry
Residential Electricity
Food/Textile/
Paper
Chemistry/
Metallurgy
Machines
Construction
Tertiary
Others
Industry consumption by sector 2005

ORDECSYS
2005-2050
Inputs
2. Evolution of useful demands and
of imported energy prices
Future growth rate, SECO Population increase, OFS

ORDECSYS
2005-2050
Inputs
3. Catalogue of existing and future technologies
Investment cost : 1500 MCHF/GW
O&M costs : 40 MCHF/GW/year
Lifetime : 30 years
Emissions : 0 tCO2/PJ
Upper-bound : Suisse.Eole

ORDECSYS
Results

10
9.5
9
8.5
8
7.5
7
6.5
6
5.5
5
12:00 16:00 20:00 0:00 4:00 8:00 12:00
Fig. 5. Simulation results with δ = 0.007. Although the tracking parameter
does not satisfy the conditions of Theorem 3.2, convergence still occurs.
10
9.5
9
8.5
8
7.5
7
6.5
6
5.5
ORDECSYS
»8
7.5
The 7
TOU Project - An overview
6.5
6
5.5
5
12:00 16:00 20:00 0:00 4:00 8:00 12:00
August 15 − 16, 2007
Normalized power Fig. 5. Simulation results with δ = 0.007. Although the tracking parameter
does not satisfy the conditions of Theorem 3.2, convergence still occurs.
10
9.5
9
Normalized power (kW) Fig. 6. Simulation results with δ = 0.003. At this point the tracking
8.5
8
7.5
7
6.5
6
5.5
5
12:00 16:00 20:00 0:00 4:00 8:00 12:00
August 15 − 16, 2007
parameter is small enough that the negotiation process does not converge.
PEVs charge for less time than others. As a consequence,
total demand ramps down at the beginning of the charging
interval, and ramps up at the end.
8
7.5
7
6.5
6
5.5
5
12:00 16:00 20:00 0:00 4:00 8:00 12:00
August 15 − 16, 2007
Normalized power Fig. 7. Converged Nash equilibrium for a heterogeneous population of
PEVs with δ = 0.015.
APPENDIX
The proof of Theorem 3.3 proceeds by considering, with-out
loss of generality, adjacent time instants t and s = t+1.
Local charging controls (!unt
, !unt
+1), that are optimal with
Normalized power (kW)
respect to u and xnt
, can be decomposed as !unt
= bn,∗−an,∗
and !unt
+1 = bn,∗ + an,∗ respectively. It is possible to show
that
an,∗ = arginf
an∈Sbn,∗
"#
an −
1
2
(ut+1 − ut)
+
1
4δ
$
p(dt+1 + ut+1) − p(dt + ut)
%&2'
with Sbn,∗ ! {an;−bn,∗ ≤ an ≤ bn,∗}.
Relationship (11a) can be established by contradiction.
If (11a) were not true, then it can be shown that an,∗ <
12
August 15 − 16, 2007
(ut+1 − ut), implying that !unt
+1 − !unt
< ut+1 − ut for all
n, and hence that
avg(!u
t+1) − avg(!u
t) < ut+1 − ut
where !u
10
9.5
≡
(
!un; 1 ≤ n < ∞
)
. This, however, conflicts with
the fact that 9
{un;!n < ∞} is a Nash equilibrium with respect
to u, see Theorem 2.1. Hence a contradiction.
kW)
12:00 5
Normalized power (kW)
Fig. 7. PEVs with The proof loss Demand response can flatten the load curve through iterative
negotiation processes (modelled via mean field games)
Ma, Callaway & Hiskens, 2007

ORDECSYS
Global models of TOU pricing reveals how to price
electricity based on measured elasticities
Supply Demand

ORDECSYS
0.6"
0.5"
0.4"
0.3"
0.2"
0.1"
0"
2005" 2010" 2015" 2020" 2025" 2030" 2035" 2040" 2045" 2050"
NEP"
NEP"-"CO2"
NEP"+"DR"
NEP"+"V2G"
NEP"+"DR"+"V2G"
0.7"
0.6"
0.5"
0.4"
0.3"
0.2"
0.1"
0"
2005" 2010" 2015" 2020" 2025" 2030" 2035" 2040" 2045" 2050"
NEP"
NEP"."CO2"
NEP"+"DR"
NEP"+"V2G"
NEP"+"DR"+"V2G"
Photovoltaics
Wind turbines
Demand response tends to delay
investments in renewables by allowing
demand to better match existing
production facilities’ constraints.

ORDECSYS
0.6"
0.5"
0.4"
0.3"
0.2"
0.1"
0"
2005" 2010" 2015" 2020" 2025" 2030" 2035" 2040" 2045" 2050"
NEP"
NEP"-"CO2"
NEP"+"DR"
NEP"+"V2G"
NEP"+"DR"+"V2G"
0.7"
0.6"
0.5"
0.4"
0.3"
0.2"
0.1"
0"
2005" 2010" 2015" 2020" 2025" 2030" 2035" 2040" 2045" 2050"
NEP"
NEP"."CO2"
NEP"+"DR"
NEP"+"V2G"
NEP"+"DR"+"V2G"
Photovoltaics
Wind turbines
Demand response tends to delay
investments in renewables by allowing
demand to better match existing
production facilities’ constraints.
However, when combining DR with
V2G possibilities*, investments in
intermittent renewables are
encouraged.
*Dual use of electric vehicles batteries: Vehicle to Grid.

ORDECSYS
24#
22#
20#
18#
16#
14#
12#
2010# 2015# 2020# 2025# 2030# 2035# 2040# 2045# 2050#
NEP#
NEP#-#CO2#
NEP#+#DR#
NEP#+#V2G#
NEP#+#DR#+#V2G#
Demand response tends decrease
the need for imports, by allowing assets
to be more efficiently managed.

ORDECSYS
24#
22#
20#
18#
16#
14#
12#
2010# 2015# 2020# 2025# 2030# 2035# 2040# 2045# 2050#
NEP#
NEP#-#CO2#
NEP#+#DR#
NEP#+#V2G#
NEP#+#DR#+#V2G#
Demand response tends decrease
the need for imports, by allowing assets
to be more efficiently managed.
However, when combined with V2G
possibilities, imports raise due to the
electricity demand stemming from
electric vehicles.

ORDECSYS
0.3"
0.25"
0.2"
0.15"
0.1"
0.05"
0"
Demand-response allows the energy system to
dynamically adapt to changing weather conditions
Scenario based on 2011's weather data
WN! WP1! WM! WP2! SN! SP1! SM! SP2! IN! IP1! IM! IP2!

ORDECSYS
0.3"
0.25"
0.2"
0.15"
0.1"
0.05"
0"
Scenario based on 2012’s weather data

ORDECSYS
0.3"
0.25"
0.2"
0.15"
0.1"
0.05"
0"
Scenario based on 2013’s weather data

1. The effects of demand-response and storage can be assessed through ETEMSmartGrid
2. Suisse-Romande’s households have a positive view of EVs and of DR mechanisms
3. EVs and DR can be exploited for a faster integration of renewables
4. Stochastic weather scenarios’ impact on DR and renewables has been studied
ORDECSYS
Conclusions
Perspectives
1. Integration of electricity network contraints, e.g. to define zonal pricing schemes (in progress)
2. Load shedding
3. Evaluation of the repercussion of an energy/climate policy on the value chain

Thanks!
christopher.andrey@ordecsys.com
ORDECSYS
Christopher Andrey
2014

The TOU project - Test 1

More Related Content

Recently uploaded (20)

Featured (20)

The TOU project - Test 1