Conceptual Design of Chemical Plants by Multi …pselab.chem.polimi.it/wp-content/uploads/2016/12/Sepiacci-Thesis... · Multi-Objective Optimization of Economic and Environmental

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


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


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


Process modeling and simulation

5



Pro

fita

bili

ty

Pollution



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.


Economic sustainability

7

Economic SustainabilityUnder Market Uncertainty

Economic Objective Function

Reference Component

Selection

Time Series Analysis

Time Delay

Correlation

Econometric Modeling


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


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


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


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


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


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

?


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


Multi-objective optimization

15



Pro

fita

bili

ty

Pollution


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


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


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


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


% Negative Cumulated DEP4

*Cumulated PEI does not depend on market volatility.


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).


Thank you for your attention

Conceptual Design of Chemical Plants by Multi …pselab.chem.polimi.it/wp-content/uploads/2016/12/Sepiacci-Thesis... · Multi-Objective Optimization of Economic and Environmental

Documents