semi-dispatchable generation with wind-photovoltaic

ADVANCED ENERGY MANAGEMENT TO
AVOID POWER UNCERTAINTY AND
FREQUENCY DISTURBANCE FROM
HYBRID SYSTEMS
Fernanda Alvarez-Mendozaa, César Angeles-Camachoa, Peder Bacherb, Henrik Madsenb
a
Instituto de Ingeniería UNAM,
b
DTU Compute
Content
 Generation tendencies
 Methodology for hybrid systems
 Hybrid System Modelling
 Forecast Model
 PV and WT Model
 FC and Electrolyzer Model
 Advance Energy Management System
 Study Cases
 Management Subjected to FC
 Management Subjected to Load Curve
 Discussion
 Conclusions
Generation tendencies
 The hybrid energy systems go
from small systems designs,
which supply one or many
households, to very large
isolated networks.
 Every day, Hybrid systems are
more common in distributed
systems, changing the way that
the network is designed and
operated.
Photovoltaic system at the
UNAM Renewable
energies institute (UNAM
Temixco)
Methodology for hybrid systems
• Wind power
• PV
Distributed • PEMFC
Generators • Micro-Hydro
Forecast
Data
Historic Data
+Storage H2
+ Generation
H2
PEMFC
PV
Interaction
between
them
Wind power
µhydro
Hybrid
Systems
New
Technologies
Lego Game
• Modeling and simulating the hybrid systems, based on renewable energies, in
multiphase frame of reference power networks, will allow a more efficient and secure
use of clean energy, which will permit a greater rate of growth in power generation
and greater reduction of pollutants in the environment.
Hybrid System Modelling
Forecast
 Energy forecasting models have become an important tool for dispatch planning
and market operation.
 A recursive least square (RLS) with forgetting factor method has been applied to
forecast wind power and radiation in a short-term range of one hour ahead with
a 10 minutes data resolution.
 The k-step AR(𝜌) model with the standard notation using X as the regressor vector
can be written like
𝑌𝑡+𝑘 = 𝑋𝑡𝑇 θt + εt+k
 RLS with a forgetting factor is based on the AR process and allows the parameter
vector 𝜃 to change over time.
Hybrid System Modelling
PV and WT model
 A photovoltaic (PV) system is composed by several photovoltaic solar cells; each
cell can generate 1 - 2 W, depending on the material used.
 The parameter 𝑅𝑠 is the cell resistance, which represents the semiconductor
material resistance. Moreover, 𝑅𝑝 is the cell shunt resistance; this stands for losses
caused by the leakage current.
𝐼 = 𝐼𝑝𝑣 − 𝐼𝑜 exp
𝑉+𝑅𝑠 𝐼
𝑉𝑡 𝛼
𝐼𝑜 =
−1 −
𝑉+𝑅𝑠 𝐼
𝑅𝑝
∵ 𝑉𝑡 =
𝑘𝑝𝑣 𝑇
Cell equivalent circuit for a
crystalline type
𝑞
𝐼𝑠𝑐𝑛 +𝐾𝑖 Δ𝑇
exp
𝑉𝑜𝑐𝑛 +𝐾𝑣 𝛥𝑇
𝛼𝑉𝑡
−1
 The current 𝐼𝑝𝑣 is linear with respect to the temperature
𝑅𝑝 + 𝑅𝑠
𝐺
𝐼𝑝𝑣 = (𝐼𝑝𝑣𝑛 + 𝐾𝑖 Δ𝑇)
∵ 𝐼𝑝𝑣𝑛 =
𝐼𝑠𝑐𝑛
𝐺𝑛
𝑅𝑝
where 𝐺 is the actual radiation and 𝐺𝑛 is the nominal radiation.
𝑘𝑝𝑣 is Boltzmann’s constant
𝑞 is the electron’s charge
𝑇 is the ambient temperature (°K)
𝐼𝑜 is the saturation current
𝛼 is the diode quality factor (1 for germanium and 2 for silicon)
Vmp : voltage at the maximum power point
Kv : Open-circuit voltage/temperature coefficient
KI: the short circuit current/temperature coefficient
Pmax,e : maximum experimental peak output power
Hybrid System Modelling
PV and WT model
 The relation between 𝑅𝑆 and 𝑅𝑝 may be found by making 𝑃𝑚𝑎𝑥,𝑚 = 𝑉𝑚𝑝 𝐼𝑚𝑝 = 𝑃𝑚𝑎𝑥,𝑒
 By solving the resulting equation for 𝑅𝑠 , we obtain
𝑃𝑚𝑎𝑥,𝑚 = 𝑉𝑚𝑝
𝑅𝑝 =
𝑉𝑚𝑝 + 𝑅𝑠 𝐼𝑚𝑝
𝑉𝑚𝑝 + 𝑅𝑠 𝐼𝑚𝑝
𝐼𝑝𝑣 − 𝐼𝑜 exp
−1 −
𝑉𝑡 𝛼
𝑅𝑝
𝑉𝑚𝑝 𝑉𝑚𝑝 +𝑅𝑠𝐼𝑚𝑝
𝑉𝑚𝑝 𝐼𝑝𝑣 −𝑉𝑚𝑝 𝐼𝑜 exp
𝑉𝑚𝑝 +𝑅𝑠 𝐼𝑚𝑝
𝑉𝑡 𝛼
+𝑉𝑚𝑝 𝐼𝑜 −𝑃𝑚𝑎𝑥,𝑒
𝑅𝑝,𝑖𝑛𝑖 = 𝐼
𝑉𝑚𝑝
𝑠𝑐𝑛 −𝐼𝑚𝑝
−
Vmp : voltage at the maximum power
point
Pmax,e : maximum experimental peak
output power
𝑉𝑜𝑐𝑛 −𝑉𝑚𝑝
𝐼𝑚𝑝
 The wind turbine model is based on the well-known static relation
𝑃 : Extracted power in watts
1
𝜌: Air density in kg/m
3
𝐴: Area swept by the rotor in m
𝑃𝑤 = 𝜌𝐴𝑐𝑝 𝑣𝑤
𝑐 : Power coefficient
2
𝑣 : Wind speed upstream of the rotor in m/s
𝑤
3
2
𝑝
𝑤
Hybrid System Modelling
Fuel Cell and Electrolyzer Model

The rate usage on the hydrogen fuel cell, which is associated with the current variable (𝐼 =
electric power generated is given as
𝑃𝑒
𝐻2𝑢𝑠𝑎𝑔𝑒 =
2𝑉𝑐 𝐹
𝑃𝑒
𝑉𝑐
), the
By using the unit moles/s for 𝐻2𝑢𝑠𝑎𝑔𝑒 , 𝑃𝑒 is the power per cell, 𝑉𝑐 represents the voltage cell and 𝐹 is the
Faraday constant (96485 C/mol).

The electrolyzer model is achieved by Faraday’s law applied to the hydrogen, which can be
combined with the ideal gas law. It is possible to compute the hydrogen volume (𝑉𝐻 2 ) in litres
generated at certain power (𝑃𝑒𝑙 ) as
𝑉𝐻2 =
𝑃𝑒𝑙 𝑅𝑇𝑡
2𝐹𝑝𝑉𝑐𝑒𝑙
𝑅 is the universal gas constant, 𝑡 is the time in seconds, 𝑇 is the temperature in °K, 𝑉𝑐 𝑒𝑙 is the voltage
cell in the electrolyzer cell and 𝑝 is the pressure in atm.
Advance Energy Management System
 An energy management system (EMS) is a system of
computer-aided tools used by operators of electric utility
grids to monitor, control, and optimize the performance
of the generation and/or transmission system.
From: research.ece.ncsu.edu
 The AEMS was developed implementing model
predictive control (MPC), alongside a time series
analysis algorithm to compute forecast, as well as the
models of the elements that integrate the HS.
Advance Energy Management System


u(t+k|t)
Many strategies can be applied in the MPC relying on the cost function
of the system.
u(t)
The future outputs for a determined horizon 𝑁 (Prediction horizon) are
predicted at each instant 𝑡.

The forecasted outputs for 𝑘 = 1, … , 𝑁 depend on the known values up to
instant 𝑡 and on the future control signal 𝑢𝑡+𝑘|𝑡 .

The set of future control signals is calculated by optimizing a determined
criterion in order to keep the process as close as possible to the
𝑟𝑒𝑓
reference trajectory 𝑃𝑡+𝑘|𝑡 .
^y(t+k|t)
y(t)
Future
control
Predicted
outputs
N
t-1
t
t+1
Cost
Function
Future
Errors
Reference
Trajectory
...
t+k
...
Constrains
Optimizer
Predicted Outputs
t+N
Past Inputs
& Outputs
Model
Advance Energy Management System
HS Generation as linear as possible
Model Predictive Control
Cost
Function
Future
Errors
Reference
Trajectory
Constrains
Optimizer
Past Inputs
& Outputs
Model
Or
Generation tracking final user load
Predicted Outputs
Or
Final user decision
Study Case
 Two study cases were analyzed.
 HS is integrated by:
•
5 kW SWT Iskra
•
2.4 kW fuel cell (FC) PEMFC
•
3 kW PV system KC200GT
•
Hydrogen storage of 16,500 liters
•
Electrolyzer with a hydrogen production rate of 18.8 SLPM
 The SWT and PV output active power will be used to produce hydrogen or send it to the grid.
 Hydrogen is stored and supplied to the FC to generate active power as required.
 The HS controller is the decision maker based on the AEMS developed, which takes into
account the final user needs without disturbing the distribution network when connected.
 The AEMS is determined in the cost function and constraints applied in the MPC. The cost
function contemplates the electric power of each element including the stored power.
Study Case
Management Subjected to FC
 The trajectory will be adjusted to maintain
an HS output power equal to the FC
𝑛𝑜𝑚
nominal output power (𝑃𝑓𝑐
) and
managed by the MPC as follow.
 Some constraints consider the slope as to
avoid abrupt power changes.
 By using the constraints, we can also
consider the level of hydrogen for
maintaining a minimum level in case of
contingency.
Study Case
Management Subjected to FC
Study Case
Management Subjected to Load Curve
 The power trajectory is managed by taking into
account the load curve of the final user (A
Mexican mini-store).
 The constrains consider the physical limitations of
the FC and electrolyzer.
 The dynamic behavior of the FC and Electrolyzer
are compute and implemented.
Study Case
Management Subjected to Load Curve
Conclusions
 The designed hybrid system, using AEMS algorithm and forecast
implementation, allows adjusting the output power of the system to not
disturb the original characteristics of the network.
 The implementation of forecasting in the HS concedes the benefit of prior
planning of power dispatch to inform the ISO and not disturb the frequency
of the network when connected as DG.
 The implemented AEMS proves to be capable of making the HS generation
from a non-dispatchable into a semi-dispatchable energy source.
 In both cases, it’s noticeable the HS flexibility in adjusting the output power
to the final user needs.
 The HS power generation can be stored for later use or injected to the grid
if a market is considered.
 Results show that the system can work with or without “backup” from the
grid power depending on the size of the HS and the needs of the final user.
Thanks for your attention!!
Fernanda Alvarez-Mendoza
[email protected]
[email protected]
Hybrid System Modelling
Forecast
 For the weighted least squares estimator, the weighted estimation is calculated
as
T
T
θt = 𝜃t−1 + R−1
∵ R t = λR t−1 + Xt+k Xt−k
t X t−k Yt − X t−k θt−1
where 𝑋𝑡 is the regressor vector
𝜃𝑡 is the coefficient vector and
𝑌𝑡 is the dependent variable at time 𝑡.
 The parameter 𝜆 is the forgetting factor, describing how fast historical data are
weighted down and taking values in the range of 0.90 to 0.995.
 The weights are equal to ω Δt = λΔt, where Δ𝑡 is the age of the data.