Conceptual Design of Chemical Plants by Multi...
-
Upload
hoangtuyen -
Category
Documents
-
view
215 -
download
0
Transcript of Conceptual Design of Chemical Plants by Multi...
Conceptual Design of Chemical Plants by
Multi-Objective Optimization of Economic
and Environmental Criteria
Relatore: Prof. Davide MANCATesi di
Piernico SEPIACCI matr. 837275
Anno Accademico 2015-2016
SCUOLA DI INGEGNERIA INDUSTRIALE E DELL’INFORMAZIONE
DIPARTIMENTO DI CHIMICA, MATERIALI E INGEGNERIA CHIMICA «GIULIO NATTA»
LAUREA MAGISTRALE IN INGEGNERIA CHIMICA
Piernico Sepiacci – Milan, 21 December 2016
Definition of sustainability
A sustainable product or process:
• constraints resource consumption and waste generation to an acceptable level;
• makes a positive contribution to the satisfaction of human needs;
• provides enduring economic value to the business enterprise.
2
Economy Society
Environment
Sustainable Development
Economic System
Goods and Services
Materials and Energy
Emissions and Wastes
Piernico Sepiacci – Milan, 21 December 2016
Definition of sustainability
A sustainable product or process:
• constraints resource consumption and waste generation to an acceptable level;
• makes a positive contribution to the satisfaction of human needs;
• provides enduring economic value to the business enterprise.
3
Economy
Environment
Sustainable Development
Economic System
Materials and Energy
Emissions and Wastes
Piernico Sepiacci – Milan, 21 December 2016
Structure of the work
4
Economic SustainabilityUnder Market Uncertainty
Economic Objective Function
Environmental Sustainabilityby Application of WAR Algorithm
Environmental Objective Function
Process Modelingand Simulation
Multi-Objective Optimization
Reference Component
Selection
Time Series Analysis
Time Delay
Correlation
Econometric Modeling
Energy Generation
Chemical Process
( )ep
outI ( )cp
outI
( )cp
inI( )ep
inI
( )cp
weI( )ep
weI
Impact Indicators Selection
Impact Specific to Chemical k
kPro
fita
bili
ty
Pollution
Statistical Analysis of Pareto Fronts on Several Forecast Scenarios
HTPI HTPE TTP ATP
GWP PCOP AP ODP
Piernico Sepiacci – Milan, 21 December 2016
Process modeling and simulation
5
Process Modelingand Simulation
Multi-Objective Optimization
Pro
fita
bili
ty
Pollution
Statistical Analysis of Pareto Fronts on Several Forecast Scenarios
Piernico Sepiacci – Milan, 21 December 2016
The cumene manufacturing process
6
Reactions Kinetics
1) Cumene reaction
2) DIPB reaction
3) Transalkylation
3 6 6 6 9 12C H C H C H
3 6 9 12 12 18C H C H C H
12 18 6 6 9 122C H C H C H
7
1 2.8 10 exp 104,181/ ( ) B Pr RT C C
9
2 2.32 10 exp 146,774 / ( ) C Pr RT C C
8
3,
9 2
3,
2.529 10 exp 100,000 / ( )
3.877 10 exp 127,240 / ( )
f B D
b C
r RT x x
r RT x
Pathak, A. S., Agarwal, S., Gera, V., & Kaistha, N. (2011). Design and control of a vapor-phase conventional process and reactive distillation process for cumene production. Industrial & Engineering Chemistry Research, 50(6), 3312-3326.
Piernico Sepiacci – Milan, 21 December 2016
Economic sustainability
7
Economic SustainabilityUnder Market Uncertainty
Economic Objective Function
Reference Component
Selection
Time Series Analysis
Time Delay
Correlation
Econometric Modeling
Piernico Sepiacci – Milan, 21 December 2016
The feasibility assessment of chemical plants traditionally follows the economic guidelines suggested
in the Conceptual Design of Chemical Processes by Douglas (1988).
Hierarchy of decisions
1. Batch vs. Continuous;
2. Input-Output Structure of the flowsheet;
3. Recycle structure of the flowsheet;
4. General structure of the separation system;
5. Heat exchanger networks.
Conceptual design
8
-1,50E+08
-1,00E+08
-5,00E+07
0,00E+00
5,00E+07
1,00E+08
1,50E+08
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Cu
mu
late
d E
P4
[U
SD]
Time [mo]
20
40
60
80
100
120
140
160
180
200
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Pri
ce [
USD
/km
ol]
Time [mo]
Cumene Raw materials
2 ( ) ( )EP Product Price Raw Materials Cost
3 2 ( )EP EP Reactor Cost CAPEX OPEX
4 3 ( )EP EP Separation Cost CAPEX OPEX
1
4 4 4nMonths
t
t
Cumulated EP EP nMonths EP
Piernico Sepiacci – Milan, 21 December 2016
Methodology
1. Selection of a suitable reference component, which must be:
• chosen according to the market field of the chemical plant;
• a key component for either the process or the sector where the plant operates.
For the O&G sector and petrochemical industry, a good candidate for the reference component is crude oil.
2. Definition of the sampling time and time horizon of the economic assessment;
3. Identification of an econometric model for the reference component;
4. Identification of an econometric model for the raw materials and (by)products;
5. Identification of an econometric model for the utilities;
6. Use of the identified econometric models to determine the economic impact of the designed
plant in terms of Dynamic Economic Potentials (DEPs).
Predictive Conceptual Design (PCD)
9
Piernico Sepiacci – Milan, 21 December 2016
Time series analysis
10
Use of moving-averaged valuesMoving average allows eliminating most of the high-frequency fluctuations and catching the price trend.
(Auto)correlation analysis(Auto)correlograms report the (auto)correlation index of the time series X and Y based on the time lag between them.
Identification of the candidate econometric model
Evaluation of the adaptive parametersIt requires a linear regression procedure that minimizes the sum of square errors between real and model prices.
1
1
0
1 n
t
i
m Yn
Pri
ce s
eri
es
Time
Real data Moving-averaged data
cov ,corr ,
var var
X YX Y
X Y
cov ,corr ,
var var
t t j
t t j
t t j
Y YY Y
Y Y
0
1
0 1 2 3 4 5 6 7
Time lag
Correlation index
0 1 1 2 2 1 1 2 2... ...t t t q t q t t p t pY a a X a X a X bY b Y b Y
0 1 1 2 2 ...t t t q t qX a a X a X a X
Autoregressive model
Autoregressive Distributed Lag model
2
,1
minnMonths
t ta b
t
Y Y
Pri
ce s
eri
es
Time
Piernico Sepiacci – Milan, 21 December 2016
Econometric models
11
20
40
60
80
100
120
140
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
WTI
pri
ce [
USD
/bb
l]
10
30
50
70
90
110
130
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Be
nze
ne
pri
ce
[USD
/km
ol]
0
50
100
150
200
2013 2014 2015 2016 2017 2018
WTI
pri
ce [
USD
/bb
l]20
40
60
80
100
120
140
160
180
2013 2014 2015 2016 2017 2018
Be
nze
ne
pri
ce
[USD
/km
ol]
, 0 1 , 1 2 , 2 1Crude Oil t Crude Oil t Crude Oil t Crude Oil Crude OilP a a P a P RAND X
, 0 1 , 1 , 1 1Benzene t Crude Oil t Benzene t Benzene BenzeneP a a P b P RAND X
Commodity a0 a1 b1 σ
Crude oil 3.3207 1.8286 - 98.13% 0.0313 −0.0028
Benzene 2.6984 0.0754 0,9012 94.07% 0.0843 −0.0098
2R X
Piernico Sepiacci – Milan, 21 December 2016
Environmental sustainability
12
Environmental Sustainabilityby Application of WAR Algorithm
Environmental Objective Function
Energy Generation
Chemical Process
( )ep
outI ( )cp
outI
( )cp
inI( )ep
inI
( )cp
weI( )ep
weI
Impact Indicators Selection
Impact Specific to Chemical k
kHTPI HTPE TTP ATP
GWP PCOP AP ODP
Piernico Sepiacci – Milan, 21 December 2016
The Waste Reduction (WAR) algorithm is a tool to determine the potential environmental impact
(PEI) of a chemical process.
The Waste Reduction algorithm
13
Energy Generation Process
Chemical Manufacturing
Process
Energy Supply
( )cp
inI ( )cp
outI
( )ep
inI ( )ep
outI
( )cp
weI
( )ep
weI
Waste Energy
Waste Energy
Mass Inputs
Mass Outputs
( ) ( )cp
cp out
out j kj k
j k
I M x
( ) ( )ep-g
ep out
out j kj k
j k
I M x
k
( ) ( ) ( )tot cp ep
out out outI I I
Total rate of PEI output
PEI output from the chemical process
Where:• Mj
(out) is the output mass flow rate of stream j;• xkj is the mass fraction of chemical k in stream j.
PEI output from the energy generation
Overall PEI for chemical k
?
Piernico Sepiacci – Milan, 21 December 2016
The Waste Reduction algorithm
14
General impact category Impact category Measure of impact category
Human toxicityIngestion LD50
Inhalation/dermal OSHA PEL
Ecological toxicityAquatic toxicity Fathead minnow LC50
Terrestrial toxicity LD50
Global atmospheric impactsGlobal warming potential GWP
Ozone depletion potential ODP
Regional atmospheric impactsAcidification potential AP
Photochemical oxidation potential PCOP
Overall PEI for chemical k
Where:
• ψkls is the specific PEI of chemical k for
the impact category l;• αl is the weighing factor of the impact
category l.
s
k l kl
l
Chemical Mass flow [kg/h] [PEI/kg] [PEI/h]
Benzene 22.537 0.837 18.863
Propylene 53.473 3.838 205.251
Propane 232.293 0.155 36.010
Cumene 12.682 1.196 15.164
DIPB 0.001 7.898 0.010
NO2 3.624 2.483 8.999
CO 1.594 0.305 0.486
CO2 2275.783 0.001 2.387
SO2 0.012 0.719 0.008
Methane 0.045 0.472 0.021
k ( )
,
tot
out kI
Piernico Sepiacci – Milan, 21 December 2016
Multi-objective optimization
15
Process Modelingand Simulation
Multi-Objective Optimization
Pro
fita
bili
ty
Pollution
Statistical Analysis of Pareto Fronts on Several Forecast Scenarios
Piernico Sepiacci – Milan, 21 December 2016 16
Multi-objective optimization
General formulation:
Objective functions:
Optimization algorithm: grid-search method.
1 2, , ,
. . 0 1,2, ,
0 1,2, ,
n
j e
j i
Optimize F F ... F
s t h j ... n
g j ... n
x
x x x x
x
x
x x xl u
F
<
,
1
4 1, ,nMonths
k t k
t
Cumulated DEP4 DEP k ... nScenarios
( )tot
outCumulated PEI I nHpM nMonths
To be maximized
To be minimized
Design variable Lower bound Upper bound Step size Steps number
Reactor length [m] 4 10 1 7
Inlet temperature [°C] 300 390 5 19
Piernico Sepiacci – Milan, 21 December 2016 17
Pareto optimal solutions
These plots show the non-Pareto and Pareto optimal solutions for an arbitrarily chosen economic scenario. Each
solution corresponds to a discrete point of the grid-search domain, i.e. a plant configuration.
As reactor length and inlet temperature increase, propylene conversion increases, thus the total rate of PEI output
decreases. At the same time, capital and energy costs increase, and selectivity decreases.
-2,E+06
0,E+00
2,E+06
4,E+06
6,E+06
8,E+06
1,E+07
3,E+06 5,E+06 7,E+06 9,E+06 1,E+07 1,E+07 2,E+07
Cu
mu
late
d D
EP4
[U
SD]
Cumulated PEI [PEI]-2,E+06
0,E+00
2,E+06
4,E+06
6,E+06
8,E+06
1,E+07
3,E+06 4,E+06 5,E+06 6,E+06 7,E+06
Cu
mu
late
d D
EP4
[U
SD]
Cumulated PEI [PEI]
Environmental sustainability
Eco
no
mic
su
stai
nab
ility
Environmental optimum
Economic optimum
Configuration Inlet temperature [°C] Reactor length [m] Cumulated DEP4k [MUSD] Cumulated PEI [MPEI]
Economic optimum 365 7 8.66 7.11
Environmental optimum 390 10 −0.99 3.42
Equidistant solution 385 5 5.13 4.77
Piernico Sepiacci – Milan, 21 December 2016 18
Pareto optimal solutions over 3000 scenarios
0%
5%
10%
15%
20%
25%
-25 -20 -15 -10 -5 0 5 10 15 20 25 30 35 40 45 50 55 60 65
Fre
qu
en
cy
Cumulated DEP4 [MUSD]
Economic optimum Equidistant solution Environmental optimum
2
3
4
5
6
7
8
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Pareto set cardinal number
Cumulated PEI [MPEI]
4
6
8
10
12
14
16
18
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Pareto set cardinal number
Average Cumulated DEP4 [MUSD]
0
5
10
15
20
25
30
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Pareto set cardinal number
% Negative Cumulated DEP4
*Cumulated PEI does not depend on market volatility.
Piernico Sepiacci – Milan, 21 December 2016
Conclusions and future developments
19
Conclusions
• The economic sustainability of the cumene plant is heavily conditioned by the fluctuations of commodity and
utility prices.
• The WAR algorithm can be used in conjunction with PCD to achieve both environmental and economic
sustainability.
• Whenever a modification is proposed to improve the environmental friendliness of a process, it is useful to
question its economic viability under market uncertainty.
Future developments
• It will be worth considering new processes based on a higher number of design variables to make the
optimization procedure more compliant with real plants.
• A further development could be expanding the boundaries of the study to include the upstream and
downstream activities related to the main process.
• As far as the social attribute of sustainability is concerned, it will be worth developing practical ways to measure
social sustainability for both single-site (e.g., process synthesis) and multi-site applications (i.e. SCM/EWO).