Dr. Ing. Emanuele Martelli - Politecnico di Milano, Dipartimento di Energia Attività di ricerca del gruppo GECOS sui sistemi trigenerativi ...

1

 Attività di ricerca del gruppo GECOS sui sistemi
 trigenerativi avanzati, i multi-energy systems e
 l'efficienza energetica dei processi industriali

Dr. Ing. Emanuele Martelli
Politecnico di Milano, Dipartimento di Energia

 February 7th 2018 , Ferrara

THE DEPARTMENT OF ENERGY @ POLITECNICO,
 DIVISIONS & OUR GROUP 2

The Department of Energy at Politecnico joins researchers originally belonging to 5 divisions.
It has 130 permanent researchers and professors

 1. Chemical Technologies and Processes and NanoTechnologies Division
 2. Electrical Division
 3. Nuclear Engineering Division
 4. Thermal Engineering & Environmental Technologies Division
 5. Fluid Dynamic Machines, Propulsion & Energy Systems Division
 • Fluid-dynamics of turbomachines
 • Internal combustion engines
 • Propulsion and combustion
 • Group of Energy COnversion Systems (GECOS):
 1 Emeritus professor (prof. Macchi)
 3 Full professors (Lozza, Consonni, Chiesa)
 9 Associate/Assistant professors
 6 RTDA
 4 Post docs
 5 Temporary researchers
 10 PhD students
 www.gecos.polimi.it

MAIN RESEARCH AREAS 3

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
 COAL + MPGs TO F-T LIQUIDS + ELECTRICITY, WITH CCS
 Cequiv balances to atmosphere for F-T liquids
 OUT: photosynthesis (MPGs, soil&root C), electricity credit (2,852 tC/day)

5. ENERGY EFFICIENCY AND SYSTEMS OPTIMISATION
 IN: upstream emissions, vented at plant, fuels burned in vehicle,s (2,852 tC/day)

 transportation

 carbon vented
 1,607 tC/day

 coal upstream emissions
 photosynthesis

 credit for e.e.
 prairie grasses upstream emissions

 1,810 tC/day

 223 tC/day
 735 tC/day
 fuel for

 225 tC/day
 83 tC/day

 1,032 MWLHV
 electricity
 production

6. ORC, S-CO2 AND ADVANCED POWER CYCLES
 452 MWee

 prairie grasses
 1,607 tC/day coal
 668 MWLHV 5,328 tC/day
 2,449 MWLHV

 polygeneration
 plant char

7. FUEL CELLS
 accumulation in 53 tC/day
 soil and root carbon storage
 1,022 tC/day 4,337 tC/day
 arrows’ width proportional to C fluxes

>10 ongoing EU projects (FP7-H2020)
Tens of ongoing research contracts with industries

OPTIMIZATION OF MULTI ENERGY SYSTEMS (MES) AND CHP
 4

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)

 4

OPTIMIZATION OF MES AND CHP: ONGOING COLLABORATIONS
 5

 EPFL,
 MIT, Boston ETH, Zurich
 Lausanne

 Skoltech, Univ. of Malaga
 Moscow GECOS
 group

 DEIB, Polimi Univ. of Parma

 Univ. of
 LEAP
 Bologna

 5

SHORT-TERM SCHEDULING PROBLEM 6

 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

 Objective: minimize the Daily/Weekly Operating Cost
 24·7 24·7 24·7 24·7

 ෍ CFuel,tot,t + ෍ & , , + ෍ C − , , ∓ ෍ , 
 =1 =1 =1 =1

Indep. variables: on/off of units, load of units, storage level in each time period t

SHORT-TERM SCHEDULING PROBLEM 7

 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 cannot
 find 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”.

MILP WITH PWL APPROXIMATION OF MAPS 8

 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)
 2-D PWL approximation with the
 «triangular method»
 1-D PWL approximation
 D’Ambrosio et al. 2010. Op. Res. Letters

 Usefull effect

Bischi et al. 2014. Energy, Vol. 74

MILP PWL FOR DAILY AND WEEKLY PROBLEMS 9

 E High temperature heat
 H
 elcust
 l 7000

 e elHP,comp e
 Heat Pump, qlow,HP,comp 6000
 a llow,stor
 c compression t
 5000

 High Temperature Heat [kWh]
 t U 4000

 r elpur fAB, LT qlow,AB S S 3000

 i Aux. Boiler, LT t e 2000

 c elsold o r
 elICE r. qlow,cust
 1000
 qlow,ICE
 fICE 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
 G ICE qhigh,ICE qlow,los -1000 Hour
 r qdeg Cogenerated ICE Cogenerated GT GT Post Firing
 i elGT qlow,GT Auxiliary Boiler Downgraded Demand
 d fGT qhigh,cust
 fGT,PF,Not,cog GT+PF qhigh,GT
 Bischi et al. 2014. Electricity
 qhigh,GT,PF,Not,cog
 fAB,HT
 Aux. Boiler, HT Energy, Vol. 74 5000

 qhigh,AB 4000

 3000

Computational time: 2000

 Electric Energy [kWh]
 1000

1 day operation: < 1 sec 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
 -1000
1 week operation: < 2 min -2000

 -3000

Up to 18% primary energy saving compared -4000
 Generated ICE
 Hour
 Generated GT Consumed HP

to usual operation strategies! Purchased Sold Demand

OPTIMAL OPERATION WITH COGENERATION INCENTIVES 10

1. COGENERATION INCENTIVES: if CHP units allow to save fuel compared
to “business as usual systems”, subsides are granted (additional revenue)

Conditions to be met for ICE CHP:
 First Principle Efficiency  Primary Energy Saving (PES)
 ℎ + _ − _ 
 ≥ 75% ≥ 10%
 _ 
 Easily rearranged as linear constraints but yearly-basis!

2. SEASONAL STORAGE SYSTEMS: Underground systems (water tanks,
aquifers, etc), sorption systems (e.g., sodium sulfide), thermochemical
storage systems and H2 storage.

Challenge: operation must be optimized considering the whole year!

IDEA: Rolling-horizon algorithm (Bischi et al. 2018. Energy, in press)

ROLLING HORIZON ALGORITHM FOR OPTIMAL OPERATION WITH INCENTIVES 11

Test case: CHP system of a large hospital
Computational time: 21 hours (worst case)  applicable for day-ahead
 scheduling
 elcust
 E qdiss,low, AB 1
 fAB, LT 1 H
 l qlow,AB 1
 Aux. Boiler LT 1 e
 e qdiss,low, AB 2 a llow,stor
 fAB, LT 2 qlow,AB 2
 c Aux. Boiler LT 2 t
 t elpur U
 qdiss,low,ICE 1 S
 r S
 elsold fICE 1 qlow,ICE 1 t
 i qlow,cust e
 ICE 1 o
 c elICE 2 qhigh,ICE 1 r. r
 qhigh,diss,ICE 1 qlow,los
 qdiss,low,ICE 2 qdeg
 G
 fICE 2 qhigh,cust
 r qlow,ICE 2,
 i elICE 1 ICE 2
 qhigh,ICE 2
 d qhigh,diss,ICE 2
 fAB, HT 1

 Bischi et al. 2018. Energy (in press)
 Aux. Boiler HT 1 qhigh,AB 1
 qhigh,diss,AB 1
 fAB, HT 2
 Aux. Boiler HT 2
 qhigh,diss,AB 2 qhigh,AB 2

About 7% higher revenue (yearly basis) compared to the weekly optimal operation.

OPTIMAL OPERATION OF MES AND CHP SYSTEMS UNDER
 12
FORECAST UNCERTAINTY
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

Minimizing the daily operating cost in the most probable scenarios

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)

OPTIMAL OPERATION OF MES UNDER FORECAST UNCERTAINTY:
ROBUST MILP 13

 Pel [kW]

 Time [h]

OPTIMAL DESIGN OF MES AND CHP SYSTEMS 14

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

DESIGN OPTIMIZATION APPROACHES 15

 Linearization without decomposition Design-scheduling decompositiom
(Gabrielli et al. 2018, Applied Energy, in press) (Elsido et al., 2017. Energy Vol. 121)
(Zatti et al, 2017, Comp. Aided Chem. Eng., 40)
 Upper level (evolutionary alg.): optimizes
Units’ selection, sizes and operation optimized in design variables (selection/sizes)
a single large scale MILP Lower level: MILP scheduling problem

Advantages: Avantages:
• Linear problem (computationally efficient)
 • Size effects accounted for on both
• Global optimality guarantee performance and costs
• Uncertainty can be rigorously handled • Possibility of considering many operating
 periods solved in sequence
Disadvantages: Disadvantages:
 Scale effects on investment costs must be
 linearized  Slow convergence rate of evolutionary
 algorithms
 Size effects on efficiency must be
 approximated  No optimality guarantee

OPTIMAL DESIGN OF MES: DESIGN OF DH NETWORKS FOR A CITY 16

 Deterministic MILP solution Stochastic MILP solution
 CAPEX+OPEX= 21,92 M€ CAPEX + OPEX= 18,22 M€
 boiler

 site 2 site 2 heat pump

 1,4 MW 1,4 MW gas turbine

 1,4 MW 1,4 MW HS heat storage

 32,6 MWh HS 41,5 MWh thermal connection
 HS

 Q23,max = 0,9 kW Q12,max = 1,8 MW Q23,max = 1,3 kW
 Q12,max = 0,9 MW

 site 1 site 3 site 1 site 3

 1,4 MW 4,5 MW
 6,7 MW
 11 MW
 4,5 MW
 1,4 MW 5 MW
 Q13,max = 4,1 MW Q13,max = 2,5 MW
 HS ES
 78,5 MWh 13,7 MWh ES HS 6,2 MWh
 HS HS 78,5 MWh

Accounting for uncertainty of forecasts leads to a cost saving of about 20%!

OPTIMAL DESIGN OF MES: RETROFIT OF DH NETWORKS 17

 DH network of Brescia University of Parma Campus

 Storage
 Tipo 1
 PdC Fumi TU Linea1

 Storage
 Tipo 1
 PdC Fumi TU Linea2 GR3 Caldaie

 Storage
 Tipo 2
 Lamarmora Lamarmora
 PdC Fumi TU Linea3

 Recupero
 Calore:
 Ori Martin
 Recupero
 Calore: Altre
 P.d.C industrie
 (Verz.+Caff
 +Concesio) Caldaie
 Extra
 Solare
 Storage

 Termico
 Tipo 3

 Caldaie
 Storage
 Tipo 3
 Centrale Nord
 EFFICITY project:
Tritorno Tmandata
 Collab. with LEAP, CIDEA & SIRAM
 Utenze RTR

 Collab. with Univ. Of Brescia

LABORATORIES 18

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 transient
conditions).

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.

LAB FOR MICRO-COGENERATION (LMC) 19

 1 kWel CHP unit based on Stirling cycle

 10-20 kWel CHP units based on internal
 combustion engines

 20-30 kWel CHP units based on PEM fuel cell with
 steam reformer

 1 kWth membrane reactor test stand (metallic
 membranes for hydrogen separation and
 application to fuel processing )

 0.5 kWth Thermo-photo-voltaic (TPV) generator

 4 x 2.5 kWel SOFC CHP unit 230

 170
 30 kWel - 18 kWLHV H2 electrolyzer unit

 At steady-state and transient / cyclic conditions
 Controlled and measured fuel / electricity exchange,
 temperature, pressure and mass flow of all I/O hot/cold
 circuits.

 19

LABORATORIO DI SISTEMI SOLARI - SOLAR TECH LAB 20

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 tools
Facilities:
- PV and PVT panels with adjustable tilt angle and distance
- Thermal oil loop to test PVT panels
- Meteorological station

 20

LEAP 21

 Attività LEAP nel settore cogenerazione ed efficienza industriale
• 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.

 21

MAIN PUBLICATIONS ON MES/CHP/MICRO-GRIDS OPTIMIZATION 22

1. Gabrielli et al., 2018. Optimal design of multi-energy systems with seasonal storage. “Applied
 Energy” (in press).
2. Zatti et al., 2017. A three-stage stochastic optimization model for the design of smart energy
 districts 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 networks
 of CHP units. Energy, Vol. 121, pp. 403-426..
4. Taccari et al. 2015. Short-term planning of cogeneration power plants: a comparison between
 MINLP 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 Mixed
 Integer Linear Programming Approaches. Proceedings of ASME Turbo-Expo 2014.
6. Bischi et al., 2014. A detailed optimization model for combined cooling, heat and power system
 operation planning. Energy, Vol. 74, pp. 12-26.
7. Bischi et al. 2016. Distributed cogeneration systems optimization: multi-step and mixed integer
 linear 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-grid
 rural microgrids. Renewable Energy, Vol. 108.
10. Mazzola et al., 2016. The potential role of solid biomass for rural electrification: A techno
 economic analysis for a hybrid microgrids in India. Applied Energy, Vol. 169.

23

Thank you for your attention!

 emanuele.martelli@polimi.it
 www.gecos.polimi.it

 23

OPTIMAL DESIGN OF A SWISS ENERGY DISTRICT (WITH ETH) 24

 Possible units Selected units

Gabrielli et al. 2018, Applied Energy, in press