sen) System Dynamics
—~* Conference POLITECNICO
MILANO 1863
MUR OROUORROROROUORROROROURURORORGUORROROROUHUROROROUHORORUROORURORURRERURGUURRORUROUOORORUROUHRROROROUHRRURORRUBERORORR|
Investigating and modelling endogenous socio-economic
dynamics in long-term electricity demand forecasts for rural
contexts of developing countries
Fabio Riva ©
fabio.riva@polimi.it
Acknowledgment to:
Emanuela Colombo (Politecnico di Milano)
Elias Hartvigsson (Chalmers University of Technology
The research NEED
research CONTEXT
The global dimension of access to electricity issue in remote areas of DCs:
2030 At least 2.5 billions to be electrified (today deficit plus the
projected population growth)
*The IEA and the WB - Global Tracking Framework Report 2015
The need of sustainable electricity planning approaches
The research NEED
research PROBLEM
How to project electricity needs and consumption patterns
within long-term off-grid electricity planning
Electricity consumption is forecasted to grow fast in developing
contexts, ESPECIALLY IN RURAL CONTEXTS
High uncertainty
Unreliable predictions
Unreliable system design
? ?
if \ e e
SIZING ELECTRIC BUSINESS PLANS
CONTROL and and TARIFFS
REGULATION definition
The research NEED
research THEME
Specific Objectives:
1) to identify and conceptualise the local dimension of electricity demand and
socio-economic development nexus
to develop a simulation model able to generate projections of electricity
demand for unelectrified contexts
Value added (General objectives):
- To contribute to the cutting-edge research committed to develop appropriate, reliable
and robust Rural Electricity Plans for off-grid areas of the world
- To support NGOs, private companies, and public utilities in the engineering design and
investment plans for reliable off-grid power systems
4
State of the art
“existing solutions can’t”
Current models adopted in the literature for rural energy demand
2.3% 2.3% 1.2%
— — —
EXTRAPOLATION SYSTEM DYNAMICS vo
ARBITRARY TREY
EIXED DEMAND
74.4 % of the studies considers a constant energy demand along the overall life cycle
19.8 % of the studies estimates an exogenous constant annual rate of growth of the
demand based on past experience, monitored data, national trends, assumptions
) Uy
*Riva, F., Tognollo, A., Gardumi, F., & Colombo, E. (2018). Long-term energy planning and demand
forecast in remote areas of developing countries: Classification of case studies and insights from a
modelling perspective. Energy Strategy Reviews, 20, 71-89. DOI: 10.1016/j.esr.2018.02.006.
State of the art
the way-forward: why SD?
“the dynamics of growth and electrification are complex, involving many [endogenous]
underlying forces” (Khandker et al.)
a Market Se
yi production TS
a a FF Sp
Sg
/ / \ | Health
Income
Generating
|Activities (IGAs)
Household
economy
=
rope ais | rope ais |
Towards mathematical instruments able to tackle
a such techno-economic and social complex dynamics:
COMPLEX ISSUE SYSTEM DYNAMICS
—_ Z COMPLICATED ISSUE 6
SD model development (1) Conceptual isation J
MODEL PURPOSE: to investigate the local socio-economic complexities of the electricity-
development nexus and generate long-term projections of electricity
demand for rural contexts of developing countries
¢ Literature-based reference modes for long-term electricity demand in developing
contexts:
N aw } +i
° EOE
2010 2015 2020 2025 2030 mane ee
mer ea ae, EN oe | AS A F. Nerini, 2015 > energy demand E. Hartvigsson, 2016 > electricity
|EA, 2015 > electricity demand in sub- T growth ina village in Timor Leste load in rural Tanzania
Electricity
B 8
Residential sector electricity use
Palys)
3
1990 1995 2000
980
B. mat 2011 8 household electricity
demand in India
I
I
I
I
I
f
fl
fl
fl
I
I
fl
I
' Saharan Africa
fl
f
I
I
I
I
f
f
I
fl
I
fl
I
O
iin data aca) N
L
S: Mustonen, 2010 > electricity growth
paths for a rural village in Laos R U R A L
Feltesitastietastiactiedtentaatantentattedtadiadateedtateatadtediatadatadtedtateted
SD model development (1) Conceptual isation J
* Feedback loops of the system:
Review of the local dimension of electricity-development nexus:
6 main dynamics identified:
i. income generating activities |
ii. market production ~ ECONOMIC
Electricity 7~ x, |_ Iii. =household economy |
Bemana Qq iv. education ]
v. people habits + SOCIAL
vi. health
* Riva, F., Ahlborg, H., Hartvigsson, E., Pachauri, S., & Colombo, E. (2018). Electricity access and rural
development: Review of complex socio-economic dynamics and causal diagrams for more appropriate
energy modelling. Energy for Sustainable Development, 43, 203-223. DOI: 10.1016/j.esd.2018.02.003
8
(1) Conceptualisation |
SD model development
e.g. electricity demand <> Market production dynamics
Market
—* supply \
a a «4
Goods/services™ sm
AO sold Net revenues
Evening work yy
a time
7 4 Market 4
/ / Lo demand
x ifs Se
f ~~
/ | : Product a Averare
| innovation fos. oe
i \4 “io SIE
| (Bit f. ~
| (BI \ +R1)
Productivity ——
\ \ — Market innovation
ape
* Riva, F., Ahlborg, H., Hartvigsson, E., Pachauri, S., & Colombo, E. (2018). Electricity access and rural
development: Review of complex socio-economic dynamics and causal diagrams for more appropriate
energy modelling. Energy for Sustainable Development, 43, 203-223. DOI: 10.1016/j.esd.2018.02.003
SD model development (1) Conceptualisation |
e.g. electricity demand <> Market production dynamics
Market +
supply
Goods/services me
¥ sold Net revenues
A
/
R3) )
f
Evening work J / / ~
; — 7 \
ene Market ‘R2)
4 _ Kee) Sa
| 4 demand —
| _
\ ae
Mechanisatipn . Product Pe Average
_v innovation Po _:
/ —————— income
* Riva, F., Ahlborg, H., Hartvigsson, E., Pachauri, S., & Colombo, E. (2018). Electricity access and rural
development: Review of complex socio-economic dynamics and causal diagrams for more appropriate
energy modelling. Energy for Sustainable Development, 43, 203-223. DOI: 10.1016/j.esd.2018.02.003
SD model development
2) Formulation
* Converting feedback diagrams to level/stocks and rate/flows equations
SD modeller
POLITECNICO _~‘\
MILANO 1863
QUESTIONNAIRE
ABOUT THE
MODELS
STRUCTURE and
VARIABLES
FEEDBACK FROM
QUESTIONNAIRE’S
RESULTS
ee
<L »>
AS) CEA
\k [7 Il seme della
SN solidarieta
CEFA NGO’s experts in
rural Tanzania
||
THE MODEL
100 levels
333 auxiliary variables
153 constants
8 look-up tables
time-step = 0.25 week
Euler integration
SD model development
For each time-step dt of the chosen simulating horizon (e.g. 20 years), the model
generates values of electricity demand for a typical rural community.
N° of Households connected to MW sleveed for appliance
the micro-grid [HHs] ppliance]
; N¢° of electricity appliances for
a he connected HHs
Cumulated weekly electricity
demand [kWh]
N° of Income Generating
Activities connected to the grid
[IGAs]
Specific electric load for IGAs
[KWh/WeelyIGAs]
SD model development
* Estimation of parameters:
a) Interviews in |kondo — Matembwe, Tanzania
* Village of around 4000 * Hydroelectric plant of about 400 kw + 18 interviews to:
people and 800 households * installed in 2005 by CEFA NGO o MVC’s and CEFA’s experts
(HHs) * previously, no access to electricity in & legal fariners
* 100-120 income generating the village o local people involved in an IGA
activities (IGAs)
¢ Agriculture-based livelihood
\ y)
Y
managed by the local utility MVC
AIM: ~ to define a realistic range for some auxiliary variables
° ~~ to define a calibration range for constants
13
SD model development
* Estimation of parameters:
b) Model calibration
/ DATA. > / Optimisation >,
control
Historical time series (2005- - 115 parameters
2017) of Ikondo’s plant data:
- Parameter spaces taken
1. Connections of IGAs from the interview and
2. Connections of HHs the literature
3. Monthly electricity - Powell optimiser of
consumption of IGAs Vensim DSS
4. Monthly enone 4 payoff definitions for
consumption.o s 4 calibration variables
\ Tor: 517 data points TS XO yy
SD model development (3) Testing and Validation |
according to model purpose
*Y. Barlas, 1996
*)_D. Sterman, 2000 7 oe
- Boundary Adequacy: wdelous —=
in KONDO
ayo ete te
- Structure Assessment:
CEFA’S LOCAL GROUP MODELLING
EXPERTS AT POLITECNICO
19+1@= ? =
s x
_J - Dimensional Consistency: OO)
VENSIM UNIT EQUATIONS
direct structure CHECK INSPECTION
tests
- Parameter Assessment: > whe,
INTERVIEWS CEFA/SLOCAL —_pLant’s DATA
in KONDO EXPERTS
_s
x ()
ier
VENSIM
CALIBRATION
SD model development (3) Testing and Validation |
according to model purpose
*Y. Barlas, 1996
*)_D. Sterman, 2000
)
- Extreme conditions: NG) O)
VENSIM VENSIM
REALITY SyntheSim
CHECK
dt < % the size of the smallest time
constant (i.e. 1 week)
= Hy 5 LITERATURE
str ucture oriented - Integration Error:
behaviour tests <6 atten emia = x(e=at,S)
t s.t,. ——_@£§- _|_——— < 25%
KS
x(t = FINAL, dt)
VENSIM V x calibrated variable
TRIALS
16
SD model development (3) Testing and Validation |
according to model purpose
*Y. Barlas, 1996
*)_D. Sterman, 2000
= J ni Ay.
* . . eve x ( )
—— Behaviour Reproduction: i war
behaviour pattern MTKONDO. DOERTS. ANALYSIS
in VENSIM
tests Y
for n = 25 variables:
Xpy(t = FINAL) € [X, -Znmax] Vn
nmin nmax,
with Xn min »Xnmax the min and max values
for x, gathered during the local interviews
17
SD model development
Theil analysis
Connected HHs MAPE = 3.70%
200
159 2 UM= 1.96%
2 US= 0.22%
= US= 97.82%
0
i} 72 «144 216 288 360 432 504 576 648
Time [Week]
Historical Sinindakecd 30 revaaeneien cs
Electricity demand of HHs MAPE =20.58%
8000
6000 UM= 0.48%
: “ +s} | US= 0.00%
° 2000 ef g
oe Uc= 99.52%
0 a
0
72 144 216 288 360 432 504 576 648
(3) Testing and Validation
according to model purpose
Connected IGAs
25
UM= 0.23%
US= 0.08%
Uc= 99.69%
144 216 288 360 432 504 576 648
Time [Week]
Electricity demand of IGAs
7000
UM= 2.87%
US= 0.48%
US= 96.65%
; Time [Week]
(um < 3% Very lowbias in the mean }
(u“~ 0 Model and data show the same trends ]
72 144 216 288 360 432 504 576 648
Time [W.
MAPE = 5.36%
MAPE = 19.22%
Model tracks actual data except for a random error term with zero mean
Unsystematic error since the purpose of the model is not to study the cycles in the data
(4) Implementation
done
Future activities
13 villages to electrify
(about 5’600 HHs and 350 IGAs)
Sensitivity analysis and test of the model’s response to different policies
Application to a future CEFA’s project of a Small Hydro Power Plant (SHPP):
UTM 725.465 E; 9.001.143 N 1397.50 mas!
UTM 725.167 E; 9.001.159 N 1322.24 mas}
177.5m
Technical features of the future Ninga-SHPP power plan
76m
6.3 MW
6.0 MW
26,410,000 kWh
v . \
/ ona, ‘i
op User \
/ \
f we UME
f . \
{ evry
‘MTANGA
oo
\ * MINGA SHPP
*, /
\ trau “ENOIY / ;
% 7 Intake location
a um / Powerhouse location
as / length
3 _ situ s Penstock lengt!
* 7 Gross head
a Mechanical capacity
Electrical power
Annual energy output
Future activities (4) Implementation
ongoing
* Application to a future CEFA’s project of a Small Hydro Power Plant (SHPP):
1. Generating N Monte Carlo samples for the M (=115) parameters calibrated during
formulation > N combinations of the M parameters
2. SD model simulation
for i=1,..., N
SD, model — diffusion of el. connections, and el. appliances,
end
3. Derivation of the long-term load curves years
h
1 De sguscnnr Y
1 = prsteeetmanyset! Pecceeetesiind
2 2 eerie! Pn, = i a Pmermrsi!
7 :
N postin! — nial *, teil
4. Stochastic design and size of the SHPP, according to the simulated load curves
Conclusion
¢ Electricity demand is growing fast in remote areas of developing
countries
¢ Reliable predictions are mandatory in order to make sustainable local
electricity plans
¢ The evolution of electricity demand is a complex problem
* | demonstrated that SD is a viable and reliable modelling approach to
investigate this issue, by developing a bottom-up model through the
main phases of the modelling process:
- Conceptualisation
- Formulation
- Testing
¢ Future work will consider the implementation of the model to a Small
Hydro Power Plant project in rural Tanzania
14.
45.
Bibliography
main references
IEA, World Bank. (2015). Sustainable Energy for All 2015-Progress toward Sustainable Energy, World Bank. https://doi.org/10.1596/978-1-
4648-0690-2. Washington, DC.
Hartvigsson, E., Ahigren, E.,Ehnberg, J., Molander. (2015). S. Rural electrification through minigrids in developing countries: initial
generation capacity effect on cost-recovery. 33rd Int. Conf. Syst. Dyn. Soc., pp. 1e12. Cambridge, USA.
Riva, F., Tognollo, A., Gardumi, F., & Colombo, E. (2018). Long-term energy planning and demand forecast in remote areas of developing
countries: Classification of case studies and insights from a modelling perspective. Energy Strategy Reviews, 20, 71-89.
Randers, J. Elements of system dynamics method. Wright-Allen series. MIT press, 1980.
Wolstenholme, E. F., & Coyle, R. G. (1983). The development of system dynamics as a methodology for system description and qualitative
analysis. Journal of the Operational Research Society, 34(7), 569-581.
Riva, F., Ahlborg, H., Hartvigsson, E., Pachauri, S., & Colombo, E. (2018). Electricity access and rural development: Review of complex
socio-economic dynamics and causal diagrams for more appropriate energy modelling. Energy for Sustainable Development, 43, 203-223.
Vennix, J. A., Gubbels, J. W., Post, D., & Poppen, H. J. (1990). A structured approach to knowledge elicitation in conceptual model building.
System Dynamics Review, 6(2), 194-208.
Jordan, R. L. (2013). Incorporating endogenous demand dynamics into long-term capacity expansion power system models for Developing
countries. Doctoral dissertation, Massachusetts Institute of Technology.
Sterman, J. D. (2000). Business dynamics: Systems thinking and modeling for a complex world. London, United States: McGraw-Hill
Education.
Barlas, Yaman. (1996). Formal aspects of model validity and validation in system dynamics. System dynamics review 12.3, 183-210.
Qudrat-Ullah, H., & Seong, B. S. (2010). How to do structural validity of a system dynamics type simulation model: the case of an energy
policy model. Energy policy, 38(5), 2216-2224.
Senge, P. M., & Forrester, J. W. (1980). Tests for building confidence in system dynamics models. System dynamics, TIMS studies in
management sciences, 14, 209-228.
Oliva, R. (2003). Model calibration as a testing strategy for system dynamics models. European Journal of Operational Research, 151(3),
552-568.
Barlas, Y. (1996). Formal aspects of model validity and validation in system dynamics. System dynamics review, 12(3), 183-210.
Riva, F., Berti, L., Mandelli, S., Pendezza, J., & Colombo, E. (2017). On-field assessment of reliable electricity access scenarios through a
bottom-up approach: The case of Ninga SHPP, Tanzania. In Clean Electrical Power (ICCEP), 2017 6th International Conference on (pp. 340-
346). IEEE. DOI: 10.1109/ICCEP.2017.8004837. 22
thank you
4 your kind attention
