Dr. V.V.N.Kishore - University of Pune

Use of Technology Diffusion
Modeling to facilitate Future
Growth of Solar Energy in India
Dr. V.V.N.Kishore
Former Professor and Head
Dept. of Energy & Environment
TERI University, Delhi
The diffusion of an innovation has been traditionally defined
as “the process by which that innovation is communicated
through certain channels over time among the members of a
social system”
Innovation: any idea, object or practice that is
perceived as new (Apps to atomic energy, deo sprays
to decentralized power)
Members of a social system: Students of UoP,
Business entities, Banks, Government agencies etc.
Certain technologies/products need a “push”
because of concerns such as environmental
degradation, energy security etc.
• MNRE: ‘N’ for New
• Some MNRE programs:
NPBD
NPIC
Energy Plantations
Solar devices (cookers, water heaters etc.)
Small hydro
Wind
Biomass gasifiers
VESP
RVE
In the diffusion process, one research
finding keeps recurring: the cumulative
adoption vs time generally takes an S-shape
(Sigmoid curve)
Diffusion theory and Modelling
1
Cumulative
Adoption
Adoption
per period
0
0
1
Key: Non-adopter; Potential Adopter
A population of potential Adopters
Each potential adopter is subject to a pressure
to adopt
The pressure may increase as others adopt
0
10
20
30
40
50
60
70
80
90
100
Time

Idealised view of the adoption of a new
technology – “S Curve”
5
The exact form of each curve, including slope and
asymptote, may differ.
Steep initial slope: rapid diffusion
Gradual slope: slow diffusion
Rate of diffusion is a function of:
- Extent of economic advantage of that innovation
- Investment needed for adoption
- Degree of uncertainty/risk associated with that
innovation
- Government policies/influence of pressure
groups etc.
Fundamental diffusion model
Different models of diffusion
• External influence model [g(t)=const a]
• Internal influence model [g(t)= bN(t)]; a
variant also called Gompertz model, widely
used for technology forecasting
• Mixed influence model [g(t)= a+ b N(t)]; also
called Bass model
Diffusion models (cont.)
Flexible diffusion models:
-Floyd model
-Sharif-Kabir model
-Jeuland model
-NSRL and NUI model
-von-Bertalanffy model
Diffusion models (cont.)
•
•
•
•
•
•
Dynamic diffusion models
Multi-innovation diffusion models
Space and time diffusion models
Multistage diffusion models
Multi-adoption diffusion models
Diffusion models with influencing/change
agents
Bass Model
•
Bass Equation (slightly modified for RET applications):
In the context of RET diffusion analysis
p - policy or push factors
q - non-policy or other factors,
m - total estimated RET potential and
N(t) - represents the cumulative installations during the period t.
11
Bass Model (2)
With the initial condition of N = N0 at t = t0, the above equation can be
integrated to give:
12
Bass Model (3)
• Parameter estimation,
(a = p; b = q/m);
Mahajan 1985
250
N(t+1)
200
150
Step I:
100
y = -0.002x2 + 1.39x + 13.36
R² = 0.95
50
;
0
0
50
N(t)
100
150
200
250
N(t+1)
Poly. (N(
The values obtained from the direct curve fit indicated m close to
actual final achievement
Step II:
Forced S – Curve for the given Potential
- m was fixed as the technical given potential
- p and q obtained by minimizing the SSE for the observed and
fitted values by making a programme in Excel using the values
obtained from Step I as indicative values
13
AP
MAHA
GUJ
KAR
Model results for wind energy
T.N ( Wrong assessment of potential?)
15
Policy Variants
Policy
Andhra Pradesh
Gujarat
Karnataka
Tamil Nadu
Maharashtra
Tariff
Upto 2004, MNRE
guidelines
Rs. 1.75 with no
escalation till 1999;
2004-5; frozen at 3.37
for 5 years, later
increased to Rs. 3.5 for
10 years in 2008-09
Rs. 2.60 kWh (5 paise
annual esc base year
2002-3 till 2004-05
MNRE guidelines with 5%
escalation on 1994-95
prices fixed at Rs. 2.00
initially; however,
increased to MNRE
recommended tariff since
1994-95
MNRE guidelines with 5%
escalation on 1994-95
prices fixed at Rs. 2.00
initially increased to Rs.
3.01 gradually in 200001; the tariff was lowered
to Rs. 2.7 without
escalation in 2001-02 till
2008-09; two tariffs Rs.
2.75 for commissioned
prior to 15-5-2006- Rs.
2.9 for commissioned
between 15-5-2006 and
18-9-2008; Rs. 3.34 for
others
2% up to 2000.
Rs. 2.25 till 1996-97 but
linked to MNRE
guidelines; Rs. 3.5 fixed
in 2003 -04 with
0.15/kWh for esc for 13
years.
5% from 2001 onwards
Since 2003-4 2% of
energy – 0 to 50 Km
Rs. 3.37 for 20 years
from 2005-06
Wheeling
2% energy fed; not
allowed since 2008-09
2% energy upto 1999
4% from 2002-3
2% till and increased to
20% ; reduced to 5%
again
2% energy;
From 2009-10 onwards
4% of energy- 50 to 200
Km
Smaller investors
(1WEG): 7%
Banking
2%, 12 months till 200708 after that disallowed
Third Party Sale
Allowed for some period
Land availability
Mainly govt land
Tenure of the policy
Short term 5 – 10 years
with frequent changes
Other incentives
Moderate
Others : 10%
6 months
Reduced to 1 month
since 2009-10
Not allowed til 2004-05
and allowed after that.
Govt and private
Policy on hold for 3 years
from 1999- 2001; 20
years
High
12 months and 2%
5%, 12 months
6% of energy- above 200
Km
12 months
Allowed
Allowed since 2006-07
Always Allowed
Govt. Land
Private land
Private land
40 years lease
No specified; but
remained consistent
13 years
Moderate
Moderate to High
High
16
Non –policy Variants
•
•
•
•
•
Investment Climate
Technological advancement
Infrastructure
Learning
Others (capacity, institutional frameworks)
17
Development of CPI
•
Interpretation of parameters
-
•
identifying RET diffusion factors
Linking to the model coefficients
Composite policy index – an indicative measure of
influence of policies in a given set of environment
A) Identification of diffusion factors
B) Review of policy variants (Pr)
C) Assigning of weights to policy type (wi )
D) CPI = Sum of (wi * Pr)
• Correlating parameters obtained and CPI
1. t* vs. CPI
2. Normalised Growth Rate at the Time of Inflection (NGRTI) [(dN(t)/dt at t*/m)*100]
18
Linking CPI to Diffusion Parameters (Biogas)
NGRTI (%) vs. CPI
t* Vs. CPI
3
45
40
2.5
35
2
30
25
1.5
20
1
15
10
0.5
5
0
0
0.2
0.4
0.6
0.8
1
0.2
0.3
0.4
0.5
Maharashtra (I)
Karnataka
West Bengal
Madhya Pradesh
CPI
0.847
0.753
0.733
0.574
t*
11
18
25
26
NGRTI
1.8
2.9
2.4
1.2
Uttar Pradesh (V)
0.313
39
1.1
0.6
0.7
0.8
0.9
19
Diffusion parameters vary depending on the
subsidy instruments
All India
Wind Power
t* (in years)
16
NGRTI
0.08
Biogas
t*(in years)
27
NGRTI
0.01
-TI or t* in the case of wind is lower implying that the market based
approach could accelerate diffusion significantly.
-The value for t* for biogas is higher implying that the subsidy based,
state run programmes take a long time to complete the diffusion.
20
Why diffusion modeling for RETs?
• Almost all past programs based on linear approach
(budget allocations, spending, subsidies etc. static)
• No agreed/established method for phasing out
subsidies (riding a tiger)
• Recognition of the diffusion process can result in
phased spending, with subsidies/tax benefits etc.
focusing more on infrastructure development, market
promotion, policy reforms, pilots etc. during the initial
(gradual uptake) period
• Hence funding/subsidies can be inverse S-curve,
decreasing with increasing diffusion (probably done
intuitively, but modeling can help)
Cumulative Improved Chulha Installations in India (in Nos.)
Best Fit for p=0.01, q=0.11, m=120 million)
140
120
80
60
40
20
Observed
Fitted
Observed Improved Cookstoves Diffusion
in India (1985 - 2002)
t* = 21 years
2061
2057
2053
2049
2045
2041
2037
2033
2029
2025
2021
2017
2013
2009
2005
2001
1997
1993
1989
0
1985
(in million
Nos.)
100
The NPIC example:
-considerable
spending
-Tall claims on gains
-program closed
down suddenly,
Ics vanished fast,
except for some
efforts through
regional/local
efforts
-restarted with some
changes, but not
moving fast
22
The NPBD: Can the subsidies ever be phased out?
Biogas Installations
Biogas Budget
4000000
4000
3500000
3500
3000000
3000
in Rs Millions
in numbers
4500
2500000
2000000
1500000
2500
2000
1500
1000000
1000
500000
500
0
1981-82 1985-86 1990-91 1994-95 1998-99 2002-03 2006-07
Year
0
1980-85
1985-90
1992-97
1997-02
2002-07 2007-2012
Year
23
Solar water heating: Are we slow or fast?
p
q
m
0.000239
0.1077
70000000
R2
0.94
t*
56.6
dN(t)/dt at t*
1548838.85
JNNSM Targets
25000
20000
15000
MW
Series1
Series2
10000
Linear (Series1)
Linear (Series2)
5000
0
0
-5000
2
4
6
years
8
10
12
What after 2022?
• Will there be a market based uptake after grid
parity is achieved?
• 8 trillion kWh/yr required to achieve a per capita
consumption of 5000 kWh. If solar can supply, say
50%, the equivalent capacity will be 2000 GW,
compared with the target of 2 GW for JNNSM
• Diffusion modeling combined with potential
estimates can give an idea on how solar power
might take off. Constant revision of models with
field data can give a direction to future policies.