1
Dr. Ing. Emanuele MartelliPolitecnico di Milano, Dipartimento di Energia
Attività di ricerca del gruppo GECOS sui sistemi
trigenerativi avanzati, i multi-energy systems e
l'efficienza energetica dei processi industriali
February 7th 2018 , Ferrara
22
1. Chemical Technologies and Processes and NanoTechnologies Division2. Electrical Division3. Nuclear Engineering Division4. Thermal Engineering & Environmental Technologies Division5. Fluid Dynamic Machines, Propulsion & Energy Systems Division
• Fluid-dynamics of turbomachines• Internal combustion engines• Propulsion and combustion• Group of Energy COnversion Systems (GECOS):
www.gecos.polimi.it
The Department of Energy at Politecnico joins researchers originally belonging to 5 divisions.It has 130 permanent researchers and professors
THE DEPARTMENT OF ENERGY @ POLITECNICO,
DIVISIONS & OUR GROUP
1 Emeritus professor (prof. Macchi)3 Full professors (Lozza, Consonni, Chiesa)9 Associate/Assistant professors6 RTDA4 Post docs5 Temporary researchers10 PhD students
33MAIN RESEARCH AREAS
1. CARBON CAPTURE TECHNOLOGIES
2. RENEWABLE ENERGY SOURCES AND WASTE-TO-
ENERGY
3. ENERGY STORAGE, HYDROGEN AND SYNTHETIC
FUELS
4. MICRO-GRIDS AND MULTI-ENERGY SYSTEMS
5. ENERGY EFFICIENCY AND SYSTEMS OPTIMISATION
6. ORC, S-CO2 AND ADVANCED POWER CYCLES
7. FUEL CELLS
>10 ongoing EU projects (FP7-H2020)
Tens of ongoing research contracts with industries
Cequiv balances to atmosphere for F-T liquids OUT: photosynthesis (MPGs, soil&root C), electricity credit (2,852 tC/day)
IN: upstream emissions, vented at plant, fuels burned in vehicle,s (2,852 tC/day)
accumulation in
soil and root
1,022 tC/day
polygeneration
plant
carbon storage
4,337 tC/day
fuel fo
r
tran
sp
ort
ati
on
1,8
10 t
C/d
ay
electricity
production
452 MWee
1,6
07 t
C/d
ay
pra
irie
gra
sses u
pstr
eam
em
issio
ns
83 t
C/d
ay
co
al u
pstr
eam
em
issio
ns
225 t
C/d
ay
char
53 tC/day
coal
5,328 tC/day
2,449 MWLHV
prairie grasses
1,607 tC/day
668 MWLHV
carb
on
ven
ted
735 t
C/d
ay
arrows’ width proportional to C fluxes
COAL + MPGs TO F-T LIQUIDS + ELECTRICITY, WITH CCS
1,0
32 M
WL
HV
ph
oto
syn
thesis
cre
dit
fo
r e.e
.
223 t
C/d
ay
Cequiv balances to atmosphere for F-T liquids OUT: photosynthesis (MPGs, soil&root C), electricity credit (2,852 tC/day)
IN: upstream emissions, vented at plant, fuels burned in vehicle,s (2,852 tC/day)
accumulation in
soil and root
1,022 tC/day
polygeneration
plant
carbon storage
4,337 tC/day
fuel fo
r
tran
sp
ort
ati
on
1,8
10 t
C/d
ay
electricity
production
452 MWee
1,6
07 t
C/d
ay
pra
irie
gra
sses u
pstr
eam
em
issio
ns
83 t
C/d
ay
co
al u
pstr
eam
em
issio
ns
225 t
C/d
ay
char
53 tC/day
coal
5,328 tC/day
2,449 MWLHV
prairie grasses
1,607 tC/day
668 MWLHV
carb
on
ven
ted
735 t
C/d
ay
arrows’ width proportional to C fluxes
COAL + MPGs TO F-T LIQUIDS + ELECTRICITY, WITH CCS
1,0
32 M
WL
HV
ph
oto
syn
thesis
cre
dit
fo
r e.e
.
223 t
C/d
ay
4
4
4OPTIMIZATION OF MULTI ENERGY SYSTEMS (MES) AND CHP
Types of optimization problems associated to MES (for microgrids and DHC
networks):
1. Design/retrofit of the system («investment planning»)
2. Long-term operation planning accounting for yearly constraints
(incentives/seasonal storage)
3. Short-term scheduling (unit commitment)
4. Optimal control (dynamic models of units and networks)
5
5
5OPTIMIZATION OF MES AND CHP: ONGOING COLLABORATIONS
ETH, Zurich
GECOS
group
EPFL, Lausanne
Skoltech, Moscow
Univ. of Malaga
MIT, Boston
Univ. of Parma
Univ. of Bologna
DEIB, Polimi
LEAP
66
Objective: minimize the Daily/Weekly Operating Cost
𝑡=1
24·7
CFuel,tot,t +
𝑡=1
24·7
𝐶𝑂&𝑀,𝑡𝑜𝑡,𝑡 +
𝑡=1
24·7
C𝑠𝑡𝑎𝑟𝑡−𝑢𝑝,𝑡𝑜𝑡,𝑡 ∓
𝑡=1
24·7
𝐸𝑙𝑡𝑜𝑡,𝑡
Given:
➜ Forecast of Electricity demand profile➜ Forecast of heating and cooling demand profile➜ Forecast of production from renewables➜ Forecast of time-dependent price of electricity (sold and purchased)➜ Performance maps of the installed units➜ Operational limitations (start-up rate, ramp-up, etc) of units➜ Efficiency and Maximum capacity of storage systems
Indep. variables: on/off of units, load of units, storage level in each time period t
SHORT-TERM SCHEDULING PROBLEM
77SHORT-TERM SCHEDULING PROBLEM
Constraints:
➜ Electric energy balance constraint Ɐ t (linear)➜ Heating energy balance constraint Ɐ t (linear)➜ Cooling energy balance constraint Ɐ t (linear)➜ Start-up constraints Ɐ t, Ɐ unit (linear)➜ Ramp-up constraints Ɐ t, Ɐ unit (linear)➜ Performance maps of units Ɐ t, Ɐ unit i (nonconvex)
Nonconvex MINLP
Available MINLP optimizers cannotfind the optimal solution
Amaldi et al., 2017. Short-term planning of cogeneration energy systems via MINLP, in SIAM book: “Advances and Trends in Optimization with Engineering Applications”.
88MILP WITH PWL APPROXIMATION OF MAPS
Basic idea: conversion into MILP via linearization of the performance maps
Advantages of MILP formulations:- Guarantee on the global optimality of the solution- Super-efficient commercial MILP solvers (e.g., CPLEX, Gurobi)
Use
full
effe
ct
Bischi et al. 2014. Energy, Vol. 74
1-D PWL approximation
2-D PWL approximation with the «triangular method»
D’Ambrosio et al. 2010. Op. Res. Letters
99
Computational time:
1 day operation: < 1 sec
1 week operation: < 2 min
Up to 18% primary energy saving comparedto usual operation strategies!
MILP PWL FOR DAILY AND WEEKLY PROBLEMS
-1000
0
1000
2000
3000
4000
5000
6000
7000
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Hig
h T
em
pe
ratu
re H
eat
[kW
h]
Hour
Cogenerated ICE Cogenerated GT GT Post Firing
Auxiliary Boiler Downgraded Demand
-4000
-3000
-2000
-1000
0
1000
2000
3000
4000
5000
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Ele
ctri
c E
ne
rgy
[kW
h]
Hour
Generated ICE Generated GT Consumed HP
Purchased Sold Demand
qhigh,GT,PF,Not,cog
qlow,los
fICEqhigh,ICE
fAB, LT
fAB,HT
qlow,AB
llow,stor
fGT
qlow,ICE
qdeg
qhigh,AB
Heat Pump,
compression
qlow,HP,comp
E
l
e
c
t
r
i
c
G
r
i
d
elHP,comp
elpur
qhigh,cust
qlow,cust
elsold
elGT
fGT,PF,Not,cog
qlow,GT
qhigh,GT
U
S
e
r
ICE
H
e
a
t
S
t
o
r.
GT+PF
Aux. Boiler, LT
elcust
elICE
Aux. Boiler, HT
High temperature heat
ElectricityBischi et al. 2014. Energy, Vol. 74
1010OPTIMAL OPERATION WITH COGENERATION INCENTIVES
𝐹𝑢𝑒𝑙_𝑟𝑒𝑓 − 𝐹𝑢𝑒𝑙_𝐶𝐻𝑃
𝐹𝑢𝑒𝑙_𝑟𝑒𝑓≥ 10%
𝑄𝑡ℎ+𝑃𝑒𝑙
𝐹𝑢𝑒𝑙≥ 75%
First Principle Efficiency Primary Energy Saving (PES)
Easily rearranged as linear constraints but yearly-basis!
1. COGENERATION INCENTIVES: if CHP units allow to save fuel comparedto “business as usual systems”, subsides are granted (additional revenue)
Conditions to be met for ICE CHP:
2. SEASONAL STORAGE SYSTEMS: Underground systems (water tanks,aquifers, etc), sorption systems (e.g., sodium sulfide), thermochemicalstorage systems and H2 storage.
Challenge: operation must be optimized considering the whole year!
IDEA: Rolling-horizon algorithm (Bischi et al. 2018. Energy, in press)
1111ROLLING HORIZON ALGORITHM FOR OPTIMAL OPERATION WITH INCENTIVES
Test case: CHP system of a large hospitalComputational time: 21 hours (worst case) applicable for day-ahead
scheduling
About 7% higher revenue (yearly basis) compared to the weekly optimal operation.
fAB, LT 1
fICE 1
fAB, HT 2
elICE 2
Aux. Boiler LT 1
qlow,los
fAB, LT 2
fAB, HT 1
qlow,AB 1
llow,stor
fICE 2
qdeg
qhigh,AB 1
E
l
e
c
t
r
i
c
G
r
i
d
elpur
qhigh,cust
qlow,cust
elsold
elICE 1
qlow,ICE 2,
qhigh,ICE 2
U
S
e
r
H
e
a
t
S
t
o
r.
ICE 2
Aux. Boiler LT 2
elcust
Aux. Boiler HT 1
Aux. Boiler HT 2
ICE 1
qlow,ICE 1
qhigh,AB 2
qhigh,diss,AB 1
qhigh,diss,AB 2
qhigh,diss,ICE 1
qhigh,diss,ICE 2
qdiss,low,ICE 2
qhigh,ICE 1
qdiss,low,ICE 1
qdiss,low, AB 2qlow,AB 2
qdiss,low, AB 1
Bischi et al. 2018. Energy (in press)
1212
Given:
➜ Forecast of power production from renewables and its uncertainty➜ Forecast of energy demand profiles and its uncertainty➜ Forecast of electricity prices and its uncertainty➜ Performance maps of the installed units➜ Operational limitations (start-up rate, ramp-up, etc) of units➜ Efficiency and Maximum capacity of storage systems
Determine:
Nominal set-points: on/off of units, nominal load of units, storage man.Correction rules: how to adjust units loads during operation
OPTIMAL OPERATION OF MES AND CHP SYSTEMS UNDER
FORECAST UNCERTAINTY
Constraints: meet energy demands, technical limitations of units, performance maps, etc
EFFICITY PROJECT (LEAP-Polimi, CIDEA, CIRI-EA, CERR) co-funded by Emilia Romagna Region (POR-FESR 2014-2020)
Minimizing the daily operating cost in the most probable scenarios
1313
OPTIMAL OPERATION OF MES UNDER FORECAST UNCERTAINTY:
ROBUST MILP
Time [h]
Pel [kW]
1414
Given: A catalogue of possible units (e.g., CHP ICEs, HP GTs, boilers, heat pumps,
etc) and the list of available sizes (discrete or continous) A catalogue of heat storage systems Forecast of future energy demand profiles (whole year) of each building/site Forecast of future energy prices and their time profiles of each building/site
Determine: Which units and heat storage system to install in each district/building The sizes of the units and storage systems The required energy connections between sites/buildings
Considering: Nonlinear size effects on investment costs and efficiency On/off and part-load operation of the units
Objectives: Maximize the NPV/Minimize the energy consumption/CO2 emissions
OPTIMAL DESIGN OF MES AND CHP SYSTEMS
1515DESIGN OPTIMIZATION APPROACHES
Linearization without decomposition(Gabrielli et al. 2018, Applied Energy, in press)(Zatti et al, 2017, Comp. Aided Chem. Eng., 40)
Units’ selection, sizes and operation optimized in a single large scale MILP
Advantages:• Linear problem (computationally efficient)• Global optimality guarantee• Uncertainty can be rigorously handled
Disadvantages: Scale effects on investment costs must be
linearized Size effects on efficiency must be
approximated
Design-scheduling decompositiom(Elsido et al., 2017. Energy Vol. 121)
Upper level (evolutionary alg.): optimizes design variables (selection/sizes)Lower level: MILP scheduling problem
Avantages:
• Size effects accounted for on both performance and costs
• Possibility of considering many operating periods solved in sequence
Disadvantages:
Slow convergence rate of evolutionaryalgorithms
No optimality guarantee
1616OPTIMAL DESIGN OF MES: DESIGN OF DH NETWORKS FOR A CITY
Deterministic MILP solutionCAPEX+OPEX= 21,92 M€
site 2
Q12,max = 0,9 MW Q23,max = 0,9 kW
Q13,max = 4,1 MW
HS
1,4 MW
11MW
32,6MWh
site 1
ES
1,4 MW
78,5MWhHS
site 3
HS
1,4 MW
13,7MWh
1,4 MW
site 2
Q12,max = 1,8 MW Q23,max = 1,3 kW
Q13,max = 2,5 MW
HS
1,4 MW
6,7MW
41,5MWh
site 1
ES
1,4 MW
78,5MWhHS
site 3
HS
4,5MW
6,2MWh
4,5MW5MW
Stochastic MILP solutionCAPEX + OPEX= 18,22 M€
HS
heatpump
gasturbine
heatstorage
thermalconnection
boiler
Accounting for uncertainty of forecasts leads to a cost saving of about 20%!
1717OPTIMAL DESIGN OF MES: RETROFIT OF DH NETWORKS
GR3 Lamarmora
Caldaie Lamarmora
Recupero Calore:
Ori Martin
PdC Fumi
Recupero Calore: Altre
industrie
Caldaie Extra
P.d.C(Verz.+Caff+Concesio)
Solare Termico
Sto
rage
Ti
po
2
Sto
rage
Ti
po
3
Sto
rage
Ti
po
3
Sto
rage
Ti
po
1
Sto
rage
Ti
po
1
TU Linea3
TU Linea2
TU Linea1
PdC Fumi
PdC Fumi
Caldaie Centrale Nord
TmandataUtenze RTRTritorno
DH network of Brescia University of Parma Campus
Collab. with Univ. Of Brescia
EFFICITY project:
Collab. with LEAP, CIDEA & SIRAM
1818LABORATORIES
LAB for MICRO-COGENERATION (LMC)
Test bench for performance and emissions of small scale co- and tri-generation systems (ICE, Stirling engines, PEM, SOFC, etc with < 100 kWel, 300 kWth), electrolizers and H2 production systems (at steady-state and transientconditions).
SOLAR TECH
Facility for developing and testing photovoltaic, photovoltaic-thermal and concentration systems.
MICRO-GRIDS LAB
Facility for testing control strategies of hybrid micro-grids (solar panels, CHP engines, batteries, thermal storages, etc).
LABORATORIO ENERGIA AMBIENTE PIACENZA (LEAP)
Research and consultancy activities in the areas of energy efficiency, waste-to-energy plants, biomass-fired plants, and pollutant emissions.
19
19
19LAB FOR MICRO-COGENERATION (LMC)
1 kWel CHP unit based on Stirling cycle
10-20 kWel CHP units based on internalcombustion engines
20-30 kWel CHP units based on PEM fuel cell with steam reformer
1 kWth membrane reactor test stand (metallicmembranes for hydrogen separation and application to fuel processing )
0.5 kWth Thermo-photo-voltaic (TPV) generator
17
0
230 4 x 2.5 kWel SOFC CHP unit
At steady-state and transient / cyclic conditions Controlled and measured fuel / electricity exchange,
temperature, pressure and mass flow of all I/O hot/coldcircuits.
30 kWel - 18 kWLHV H2 electrolyzer unit
20
20
20LABORATORIO DI SISTEMI SOLARI - SOLAR TECH LAB
PV system optimization: - improvements of commercial devices/technologies (e.g., PV, PVT, inverters, etc)- development of innovative concepts/prototypes (e.g., solar-driven heat pump)- development of accurate forecasting toolsFacilities:- PV and PVT panels with adjustable tilt angle and distance- Thermal oil loop to test PVT panels- Meteorological station
21
21
21LEAP
• Ricerca nel settore MES e sistemi CHP in collaborazione con il gruppo GECOS
• Consulenza scientifica per il recupero di calore/energia da processi industriali (BP, GE, etc)
• Consulenza tecnica e procedurale per l’accesso a sistemi incentivanti sulla produzione efficiente di energia termica: CAR (Cogenerazione ad Alto Rendimento) e TEE (Titoli di Efficienza Energetica)
Audit energetico degli impianti;
Valutazioni tecnico-economiche sui meccanismi di remunerazione dell’energia;
Verifica della rispondenza alle normative di riferimento;
Validazione e supporto tecnico dei modelli per il calcolo del PES (Primary Energy Saving);
Assistenza per la presentazione dei documenti verso il GSE.• Consulenza tecnica e procedurale per il riconoscimento degli incentivi per la produzione di
energia elettrica da impianti a fonte rinnovabile: In particolare: per impianti a biomassa e impianti ibridi alimentati da rifiuti parzialmente
biodegradabili.
Attività LEAP nel settore cogenerazione ed efficienza industriale
2222MAIN PUBLICATIONS ON MES/CHP/MICRO-GRIDS OPTIMIZATION
1. Gabrielli et al., 2018. Optimal design of multi-energy systems with seasonal storage. “AppliedEnergy” (in press).
2. Zatti et al., 2017. A three-stage stochastic optimization model for the design of smart energydistricts under uncertainty. “Computer Aided Chemical Engineering” Vol. 40, pp. 2389-2394.
3. Elsido et al., 2017. Two-stage MINLP algorithm for the optimal synthesis and design of networksof CHP units. Energy, Vol. 121, pp. 403-426..
4. Taccari et al. 2015. Short-term planning of cogeneration power plants: a comparison betweenMINLP and piecewise-linear MILP formulations. Computer Aided Chem. Eng., 37, pp. 2429-2434.
5. Bischi et al., 2014. Tri-Generation Systems Optimization: Comparison of Heuristic and MixedInteger Linear Programming Approaches. Proceedings of ASME Turbo-Expo 2014.
6. Bischi et al., 2014. A detailed optimization model for combined cooling, heat and power systemoperation planning. Energy, Vol. 74, pp. 12-26.
7. Bischi et al. 2016. Distributed cogeneration systems optimization: multi-step and mixed integerlinear programming approaches. International Journal of Green Energy, Volume 13.
8. Mazzola et al., 2015. A detailed model for the optimal management of a multigood microgrid.Applied Energy, Vol. 154.
9. Mazzola et al., 2017. Assessing the value of forecast-based dispatch in the operation of off-gridrural microgrids. Renewable Energy, Vol. 108.
10. Mazzola et al., 2016. The potential role of solid biomass for rural electrification: A technoeconomic analysis for a hybrid microgrids in India. Applied Energy, Vol. 169.
2424
Gabrielli et al. 2018, Applied Energy, in press
OPTIMAL DESIGN OF A SWISS ENERGY DISTRICT (WITH ETH)
Selected unitsPossible units
