This document summarizes Giulio Vialetto's Ph.D. research on improving energy efficiency in industrial facilities through innovative energy systems and data analysis methods. The research analyzed systems combining solid oxide fuel cells with heat pumps for advanced heat recovery and reversible solid oxide cell systems for combined heat, power, and hydrogen production. Cluster analysis of energy demand data was also studied to better design energy systems and identify opportunities to reduce mismatch between demand and supply. Case studies found primary energy savings of 2-6.5% were possible depending on the production levels and improvements implemented.
HARDNESS, FRACTURE TOUGHNESS AND STRENGTH OF CERAMICS
Phd presentation
1. Ph.D. student: Giulio Vialetto
Supervisor: Prof. Marco Noro
ENERGY EFFICIENCY INTO
INDUSTRIAL FACILITIES
2. PREFACE
The aim of the research activity was to
improve the efficiency on energy generation in
industrial facilities by using both innovative
energy systems (“hardware”) and big data
methods (“software”). The idea is that if these
improvements are adopted at the same time,
efficiency would be higher compared to the
case they are adopted separately.
An energy system should improve both on
generation both on operation strategy.
5. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
SOFC (solid oxide fuel cell) converts
fuel into electricity and heat with high
efficiency. Heat is recovered from
waste gases that have a high
percentage of water (steam). If not
only sensible but also latent heat can
be recovered, energy efficiency of the
system is increased.
Air source heat pumps (ASHP) are
cheaper than ground source heat
pumps (GSHP). In some climates,
however, evaporation section may
freeze.
SOFC, ASHP – AN OVERVIEW
6. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
SOFC waste gases are mixed with inlet air into an adiabatic mixer,
increasing both temperature and absolute humidity.
The aim is to increase COP of ASHP and decrease the freezing of
evaporation section.
SOFC – ASHP INTEGRATED SYSTEM
7. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
Simulations were performed with a 50 kW nominal power SOFC and an
ASHP with 7.7 kW nominal heating capacity. Air inlet temperature varies
from –7.5 °C to 15 °C, relative humidity from 25% to 100%.
Two benchmarks are defined to evaluate the performances: COP variation
and %PES. COP variation verifies if COP of the system proposed is higher
than a traditional ASHP. %PES verifies which is the primary energy saving
of the innovative system compared with a traditional one.
SIMULATION PARAMETERS AND BENCHMARKS
𝐶𝑂𝑃𝑣𝑎𝑟𝑖𝑎𝑡𝑖𝑜𝑛 =
𝐶𝑂𝑃𝑖𝑛𝑛𝑜𝑣 ,𝑠𝑦𝑠
𝐶𝑂𝑃𝑡𝑟𝑎𝑑 ,𝑠𝑦𝑠
− 1 ∙ 100 %𝑃𝐸𝑆 = 1 −
𝑃𝐸𝑖𝑛𝑛𝑜 ,𝑠𝑦𝑠
𝑃𝐸𝑡𝑟𝑎𝑑 ,𝑠𝑦𝑠
∙ 100 = 1 −
𝐸𝑎𝑣𝑎
𝜂 𝑒𝑙𝑒
+
𝐻𝑎𝑣𝑎
𝜂 𝑏𝑜𝑖𝑙𝑒𝑟
𝐹𝑆𝑂𝐹𝐶
∙ 100
8. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
COP variation varying the external inlet air temperature for four very
different cases in terms of SOFC nominal power and air relative humidity.
RESULTS - COP VARIATION
9. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
Primary energy saving varying the external inlet air temperature for four
very different cases in terms of SOFC nominal power and air relative
humidity.
RESULTS - %PES
10. Polygeneration system – Hydrogen
production with RSOC
ALTERNATIVE ENERGY GENERATION
SYSTEM FOR INDUSTRIAL FACILITY
11. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
A reversible solid oxide cells (RSOC)
system could work as solid oxide fuel cells
(SOFC) producing energy (electricity and
heat at high temperature) or as electrolyser
(solid oxide electrolyser cells, SOEC)
where heat and electricity are used to
produce hydrogen.
It is proposed that a combined system
composed by some sub-systems working
as SOFC and some as SOEC creates a
reversible energy system where is possible
to vary H/P ratio having hydrogen as sub
product.
RSOC – AN INTRODUCTION
RSOC
HE G
12. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
Varying the ratio between RSOC working as SOFC and SOEC (nRSOC), heat
to power ratio varies too. It could cover the range of the other cogeneration
technologies.
RSOC - HEAT TO POWER VARIATION
𝑛 𝑅𝑆𝑂𝐶 =
𝑃𝑆𝑂𝐸𝐶
𝑃𝑆𝑂𝐹𝐶
𝐻
𝑃 𝑅𝑆𝑂𝐶
=
𝐻
𝑃 𝑆𝑂𝐹𝐶
−
𝐻
𝑃 𝑆𝑂𝐸𝐶
∗ 𝑛 𝑅𝑆𝑂𝐶
1 − 𝑛 𝑅𝑆𝑂𝐶
𝑃𝑆𝑂𝐹𝐶 =
1
1 − 𝑛 𝑅𝑆𝑂𝐶
∗ 𝑃𝑅𝑆𝑂𝐶
𝑃𝑆𝑂𝐸𝐶 =
𝑛 𝑅𝑆𝑂𝐶
1 − 𝑛 𝑅𝑆𝑂𝐶
∗ 𝑃𝑅𝑆𝑂𝐶
13. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
Paper production is an intensive energy consumption and it requires both
electricity and heat. A paper mill asked to analyse its energy generation
system to improve efficiency.
While working on operation data it was decided to propose an alternative
energy generation system: RSOC are proposed to improve energy
production and, when production rate is low, to produce hydrogen. The
farm has two production lines, it could work only Line 1 (Case 1), only
Line 2 (Case 2) or both of the lines (Case 1+2). Energy consumption and
also heat to power ratio vary depending on the lines working.
CASE STUDY – AN OVERVIEW
14. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
The traditional energy system (left) is improved by RSOC (right). One of
the two steam turbines (the oldest part of the system, installed in the ‘60)
could be dismissed.
ENERGY SYSTEM IMPROVEMENT PROPOSED
15. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
Adoption of RSOC could increase efficiency on energy generation: it is
estimated that if all of the production lines work (Case 1+2), it is possible to
achieve a primary energy saving (PES) of 6.5% without the production of
hydrogen. Meanwhile, if only line 1 (Case 1) or line 2 (Case 2) works,
hydrogen is produced with a flow rate of 16.14-16.86 kg/h, a PES of 2% on
energy production and a PES of 45% on hydrogen production can be
reached.
THERMODYNAMIC ANALYSIS
CASE H2 PROD. PES EN. GEN. PES H2 gen
Case 1 16.857 kg/h 2.67% 45.62%
Case 2 16.137 kg/h 2.27% 45.28%
Case 1+2 - 6.54% -
16. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
The aim of the system is not only to
increase efficiency but also to
produce hydrogen with a lower cost
compared to other technology.
A sensitive analysis on RSOC
purchase cost varying it between -
10% and 30% show that H2 cost
varies between 6-8 €/kg (whereas the
costs is 10 €/kg if it is produced by
using Proton Exchange Membrane
Electrolyser (PEMEC)).
HYDROGEN COST
18. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
Meanwhile more data on energy demands are available,
energy system are still analysed using cumulative curve
of consumption. In a case that two types of energy (for
example heat and electricity) are consumed, it is
unknown which correlations there are between them.
(Figure taken from A. Biglia, F. V. Caredda, E. Fabrizio, M. Filippi, and N. Mandas,
“Technical-economic feasibility of CHP systems in large hospitals through the Energy
Hub method: The case of Cagliari AOB,” Energy Build., vol. 147, pp. 101–112, Jul.
2017)
SIZING COGENERATION SYSTEM – AN OVERVIEW
19. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
It is proposed to use cluster
analysis to perform
clustering on energy data
demands.
The main scope is to divide
the observed data into
homogenous groups and use
them to design and size an
energy system.
CLUSTERING – AN INTRODUCTION
20. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
Two different analyses based on clustering are proposed:
• Power analysis, every observation is considered separately to define
clusters with similar values of the variables (i.e. electricity demand and H/P
ratio). This information, and how such variables vary inside the cluster,
will suggest the most suitable polygeneration technology and/or
information to design the generation system;
• Profile analysis, daily energy demand profile (not a single observation) is
defined and clustered to identify how energy demand varies during
daytime. Possible mismatching can be detected between energy demand
and energy production using energy system defined with Power analysis.
CLUSTERING AND ENERGY DATA – PROPOSED ANALYSES
21. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
A workflow is then proposed to perform cluster analysis both for power and
profile analysis. Data cleaning is necessary to clean dataset from missing
and/or bad measurement records. A MATLAB script combined with
Machine Learning toolbox was defined to perform Power and Profile
analyses.
ANALYSIS WORKFLOW
• Import dataset
• Data validation
and cleaning
DATASET
• Application of
silhouette criteria
to define number
of cluster
DEFINE
HYPERPARAMETERS
• Clustering with K-
Means
CLUSTERING
• Definition of
cluster average
curves
AVERAGE CURVES
22. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
A case study is proposed concerning an
industrial facility selling wood (timber)
window laminated, plywood, engineered
veneer, laminate, flooring and white wood.
The industrial process requires to dry wood
into kilns, and to store it into warehouses.
Electricity is used for the production
equipment, offices, lighting purpose into the
warehouses, and to charge electric forklifts.
Heat is used to produce steam for the kilns
that work at about 70 °C.
CASE STUDY – INTRODUCTION
23. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
CASE STUDY - POWER ANALYSIS
24. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
Cluster Number of observations
1 31.91 %
2 21.90 %
3 0.27 %
4 45.92 %
CASE STUDY – PROFILE ANALYSIS
25. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
On the dataset both power and
profile analyses are performed.
Firstly power analysis suggests
the most suitable cogeneration
system – micro gas turbines.
Profile analysis gives also useful
information to define operation
strategy and energy storage (in
this case heat).
CASE STUDY – PROPOSED IMPROVEMENTS
26. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
Two different TO BE
scenarios are proposed to
improve efficiency on
energy generation.
First, an improvement
only on energy generation
(microturbines) is
proposed with heat
storage.
CASE STUDY – SCENARIO TO BE 1
27. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
In scenario TO BE 2
operation strategy is
improved,
cogeneration stops
when heat storage is
not able to store
more heat: the aim
is to avoid heat
losses.
CASE STUDY – SCENARIO TO BE 2
28. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
Analysis on primary energy saving (PES) between AS IS and TO BE
scenarios is then performed. It is possible to appreciate that saving of 6 %
can be achieved. Heat storage is important to achieve this goal: the mean
heat stored level is close to 50 % covering between 4 - 5 % on total heat
demand (IC).
CASE STUDY – BENCHMARK
Scenario Primary energy Saving
AS IS 6.505 GWh -
TO BE 1 6.377 GWh 2.01 %
TO BE 2 6.137 GWh 6.00 %
𝑃𝐸 = 𝐹 +
𝐸 𝑔𝑟𝑖𝑑,𝑖𝑛 − 𝐸 𝑔𝑟𝑖𝑑,𝑜𝑢𝑡
0.434
𝐼𝑆 =
𝐻𝑠𝑡𝑜𝑟𝑒𝑑,𝑖𝑛
𝐻 𝐶𝐻𝑃
𝐼 𝐶 =
𝐻𝑠𝑡𝑜𝑟𝑒𝑑,𝑜𝑢𝑡
𝐻 𝑢𝑠𝑒𝑟
Scenario IS IC % Mean heat stored
TO BE 1 4.6 % 4.3 % 50.5 %
TO BE 2 5.7 % 4.7 % 48.9 %
29. Clustering and kNN for short-term
forecasting
BIG DATA ANALYSIS FOR ENERGY
EFFICIENCY
30. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
Clustering is proposed not only to design energy system but also to increase
their efficiency forecasting energy consumption data. Clustering is proposed
to find similar patterns of consumption and, consequently, average patterns
of consumption. These (average) patterns are then used to forecast
consumption using k-Nearest Neighbour (kNN) machine learning method.
CLUSTERING FOR FORECASTING – AN OVERVIEW
• Observation
dataset trains the
model
MODEL
TRAINING
• Observations are
used to classify the
correspondent
average curve
CURVE
CLASSIFICATION • Average curve is
used to define
forecast
FORECAST
31. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
A workflow is defined to train the model and choose its parameters
(hyperparameters). Novelties are also proposed on dataset normalisation
method and hyperparameter definition. Both of the workflows are
implemented with a MATLAB script using Machine Learning toolbox.
FORECASTING WORKFLOW
• Definition and
normalisation
• Define validation,
training and test
dataset
DEFINE
DATASET
• Define hyper
parameters of
clustering and kNN
using validation
dataset
DEFINE HYPER
PARAMETERS • Define clusters using
training dataset
TRAIN CLUSTER
MODEL
• Define kNN model
using training
dataset
TRAIN kNN
MODEL • Verify model using
test dataset
TEST MODEL
32. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
Firstly instead of
normal score, a
percentage norm is
proposed. For each
observation, average is
calculated and then
used to normalise
observation.
It is expected that this
method decreases only
scale effect on dataset.
DATA NORMALISATION
33. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
Mean absolute percentage
error (MAPE) is then
proposed to define the
optimum number of cluster
to divide the dataset. This
method is useful to predict
which would be the error on
forecasting. Number of
cluster (n) could be defined
as:
MAPE CRITERIA FOR HYPERPARAMETER DEFINITION
min(n) | MAPE(n) < (MAPE(n+1)+MAPE(n+2)+MAPE(n+3))/3min(n) | MAPE(n) < MAPE_limit
34. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
Dataset previously used for the
previous analysis was used also to
test the proposed forecast method.
Firstly, it is possible to appreciate
that MAPE criteria was able to
predict error on forecast when
training and test is performed. It is
possible to appreciate that forecast
error in some cases is about 3.5 %.
CASE STUDY - INTRODUCTION
35. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
Novelties proposed
on normalisation
(percentage norm)
decreases MAPE
error compared to
standard score.
IMPROVEMENT ON DATA NORMALISATION
36. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
Performance on
electricity (on top) and
on heat (on bottom)
demand forecast varying
observed demand (supp
ort) and forecasted
values (forecast).
CLUSTERING FOR FORECASTING
37. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
MAPE between validation dataset and test dataset. Validation dataset is able
to predict MAPE error on test dataset
MAPE BETWEEN VALIDATION AND TEST DATASET
Curve Energy
Validation dataset Test dataset
MAPE MAPE1 MAPE RMSE1 RMSE
8-4 Electricity 3.60% 2.75% 3.58% 5.15 3.82
8-4 Heat 35.41% 32.95% 34.11% 93.43 55.43
10-4 Electricity 3.71% 2.74% 3.57% 5.15 3.82
10-4 Heat 35.23% 32.7% 34.95% 93.2 54.82
10-8 Electricity 4.79% 2.9% 4.47% 5.47 3.53
10-8 Heat 36.66% 35.3% 34.12% 90.03 41.99
12-8 Electricity 4.69% 2.8% 4.47% 5.31 3.53
12-8 Heat 39 % 32.1% 37.21% 95.14 43.05
.
38. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
In this thesis improvements on energy generation both using SOFC/SOEC both data
analytics are proposed. In the first part innovation on SOFC are proposed to increase
efficiency on energy generation.
A novel heat recovery for system composed by ASHP and SOFC is proposed and
analysed. Simulations show that it is possible to increase efficiency of the system, COP
is higher when a powerful SOFC is available and when air has a high relative humidity.
Then Reversible solid oxide cells (RSOC) are proposed as flexible energy system where
it is possible to vary H/P ratio by modifying the sub-systems working as SOFC and as
SOEC. Hydrogen is produced as sub-product. RSOC is proposed to improve energy
generation into an industrial facility (paper mill) to dismiss an old steam turbine.
Primary energy saving occurs varying between 2.27 % - 6.5 %. Hydrogen could be
produced with a rate of 16 kg/h with a lower cost compared to traditional electrolyser
such as PEMEC.
CONCLUSION
39. OVERVIEW METHOD SIMULATION CONCLUSION
Energy efficiency into industrial facilities – Ph.D. student Ing. Giulio Vialetto, supervisor Prof. Marco Noro
Doctoral School of Industrial Engineering, Curriculum Energetic Engineering - XXXII Cycle – University of Padua
Data analytics then proposed to improve efficiency using energy demand data.
Clustering is used to divide dataset into homogenous groups to define which is the most
suitable energy generation technology with power analysis, profile analysis is then
used to check if energy storage occurs and/or which is the most suitable operation
strategy. Proposed methodology is then applied to an industrial case study to enhance its
energy cogeneration system. It was demonstrated that a PES of 6 % can be achieve
improving energy generation.
Clustering combined with kNN are proposed also to perform short-term forecast of
energy demand. Novelties are proposed on data normalisation to increase accuracy on
forecasting. Method proposed was then tested with a case study, MAPE on electricity
forecasting was 3.6 %. Consumption forecasting could be used to improve control on
generation, to decrease energy production when it is unnecessary.
CONCLUSION
40. PUBLISHED PAPERS
Co-Authors Journal Title
Vialetto Giulio, Noro Marco
Energy Conversion and Management
(under review)
An innovative approach to design cogeneration systems based on
big data analysis and use of clustering
Vialetto Giulio, Noro Marco Energies 2019, 12(23), 4407
Short forecasting method based on clustering and kNN:
application to an industrial facility powered by a cogenerator
Vialetto Giulio, Noro Marco
Proceedings “14th SDEWES
Conference”, Dubrovnik, 2019
An innovative approach to design cogeneration systems based on
big data analysis and use of clustering
Vialetto Giulio, Noro Marco,
Colbertaldo , Rokni Masoud
International Journal of Hydrogen
Energy, 2019, 44(19), pp. 9608-9620
Enhancement of energy generation efficiency in industrial
facilities by SOFC – SOEC systems with additional hydrogen
production
Vialetto Giulio, Noro Marco,
Rokni Masoud
Journal of Electrochemical Energy
Conversion and Storage, 2019, 16(2),
021005, Paper No: JEECS-18-1064
Studying a hybrid system based on solid oxide fuel cell combined
with an air source heat pump and with a novel heat recovery
Vialetto Giulio, Noro Marco,
Rokni Masoud
Proceedings “12th SDEWES
Conference”, Dubrovnik, 2017,
SDEWES2017.75, ISSN 1847-7178
Analysis of a cogeneration system based on solid oxide fuel cell
and air source heat pump with novel heat recovery
Vialetto Giulio, Noro Marco,
Rokni Masoud
Journal of Sustainable Development of
Energy, Water and Environment
Systems, 2017, 5(4), pp. 590-607
Thermodynamic Investigation of a Shared Cogeneration System
with Electrical Cars for Northern Europe Climate
Vialetto Giulio, Noro Marco,
Rokni Masoud
International Journal of Hydrogen
Energy, 2017 42(15), pp. 10285-10297
Combined micro-cogeneration and electric vehicle system for
household application: An energy and economic analysis in a
Northern European climate
41. INDUSTRIAL SUPPORTER
I would like to thank Mosaico S.r.L. (part of BURGO Group S.p.A.) and
Corà S.p.A. that provided useful case study for the methods proposed.