an approach for risk reduction (methodology) based on...

MKOPSC Symposium

QRA

CFD

OPT

An approach for risk reduction (methodology) based on optimizing

the facility layout and siting in fire and explosion scenarios

1

Dedy Ng,

Seungho Jung, Christian Diaz Ovalle, Richart Roman Vazquez,

M. Sam Mannan

Contents

2

Introduction Overview Motivation Research objectives

Flowchart of the Proposed Methodology Overall theme (QRA – Optimization - CFD)

Case study with Hexane distillation plant Description for case study 1st approach 2nd approach 3rd approach Evaluations with RealityLINx and FLACS

Conclusion & Future work

Overview

E-1 E-2

E-3

E-4

E-5 E-6

E-7

P-4

Process Design

Equipment Layout

+ Piping Route

Facility Layout

3

MotivationBP Texas city (2005)

TrailersPasadena (1989)

Control room

Bhopal (1984)

Affecting nearby

residential area

San Juanico disaster (1984)

Residential area

Domino effect

Separation Distance

Hazardous point vs. Occupied buildings

Residential Area

Between hazardous facilities

4www.abc.net.au/news/photos/2007/10/26/2070854.htm http://www.acusafe.com/Incidents/PasadentTexas1989/incident-pasadenatexas1989.html www.ens-newswire.com/.2004-11-30-10.asp upload.wikimedia.org/.../250px-Gaskessel_gr.jpg http://www.bbc.co.uk/threecounties/read_this/2006/12/review_of_the_year_2006.shtml (Victor Carreto reproduced)

Buncefield (2005)

Storages

Literature review

Research Objectives

Combine economic concept and safety concept

Facility layout optimization using QRA

Aim for realistic prediction! – Consider obstacle effects

– Adopt real scenarios in the model formulation

Develop a layout optimization model

Help in a decision-making for safer siting & layout

Optimization (MINLP)

GAMS

5

Methodology for 3 different approaches

FLACS-CFD code

6

Optimization

Interconnection cost

Land cost

Risk cost

Separation constraints

Layout resultsFLACS

Overpressure evaluation

QRA

Release frequency

Event tree analysis

Consequence analysis

Facility (equipment) type

Optimization

Interconnection cost

Land cost

Separation constraints

Optimization

Interconnection cost

Land cost

Risk cost

QRA

Release frequency

Event tree analysis

Consequence analysis

(1) Distance-based approach

(2) Overpressure estimation approach

(3) Integrated method based on recommended separation distance and overpressure

http://www.gexcon.co.uk/products_explo.html

Separation distance from boundary

Case Study – C6 distillation column

Description- A distillation column is used to

separate hexane and heptane

Incident outcome

Release frequency

(yr-1)

Incident outcome

probability

Incident outcomefrequency

(yr-1)

BLEVE 2.3 X 10-5 0.25 5.7 X 10-6

VCE 2.3 X 10-5 0.34 7.8 X 10-6

Flash fire 2.3 X 10-5 0.34 7.8 X 10-6

Incident outcome probability via ETA

7

AICHE/CCPS (2007) Guidelines for chemical process quantitative risk analysis

Case study – Hexane distillation plant

8

i Type Length (m-m) Boundary (m) Facility cost, FCi ($)

1 Control room (non-pressurized) 10-10 30 1,000,000

2 Administrative building 20-15 8 300,000

3 Warehouse 5-10 8 200,000

4 High pressure storage sphere 10-10 30 150,000

5 Atmospheric liquid storage tank 1 4-4 30 100,000

6 Atmospheric liquid storage tank 2 4-4 30 100,000

7 Cooling tower 20-10 30 500,000

8 Process unit 30-40 30 .

Facility I Unit interconnection cost, UICi,j ($ / m)

Re

com

me

nd

ed

se

par

atio

n

dis

tan

ce b

etw

ee

n f

acili

tie

s , D

i,j(

m) 1 0.1 0.1 10 10 10 10 10

5 2 0.1 0 0 0 0 0

5 5 3 0.1 0.1 0.1 0.1 0.1

30 60 60 4 0.1 0.1 100 0

60 60 60 10 5 0.1 100 0

60 60 60 10 4 6 100 0

30 30 30 30 30 30 7 100

30 60 60 15 5 5 30 8

1st Approach (Distance based) - formulations

Min( Land cost + ∑Interconnection costi,j )

Land cost = UL X Max(xi + 0.5 Lxi) X Max(yi + 0.5 Lyi)

Interconnection costi,j = UICi,j X di,j

di,j2 = (xi - xj)

2 + (yi - yj)2

Objective function

Constraints

"𝐿𝑒𝑓𝑡"

𝑥𝑗 ≤ 𝑥𝑖 − 𝐷𝑖 ,𝑗𝑁𝑂,𝑥 ⋁

"𝑅𝑖𝑔𝑕𝑡"

𝑥𝑗 ≥ 𝑥𝑖 + 𝐷𝑖 ,𝑗𝑁𝑂 ,𝑥 ⋁

"𝐴𝑏𝑜𝑣𝑒", "𝐷𝑜𝑤𝑛"

𝑥𝑗 ≥ 𝑥𝑖 − 𝐷𝑖 ,𝑗𝑁𝑂 ,𝑥

𝑥𝑗 ≤ 𝑥𝑖 + 𝐷𝑖 ,𝑗𝑁𝑂 ,𝑥

"𝐴𝑏𝑜𝑣𝑒"

𝑦𝑗 ≥ 𝑦𝑖 + 𝐷𝑖 ,𝑗𝑁𝑂 ,𝑦 ⋁

"𝐷𝑜𝑤𝑛"

𝑥𝑗 ≤ 𝑦𝑖 − 𝐷𝑖 ,𝑗𝑁𝑂 ,𝑦

9

𝐷𝑖 ,𝑗𝑁𝑂 ,𝑥 =

𝐿𝑥𝑖+𝐿𝑦𝑗

2+ 𝐷𝑖,𝑗 𝐷𝑖 ,𝑗

𝑁𝑂 ,𝑦=

𝐿𝑦𝑖 + 𝐿𝑦𝑗

2+ 𝐷𝑖,𝑗 where,

Non-overlapping constraints

Facility kFacility s

Region Left

Region Right

Region Above

Region Down

min,

,

min,

,min, min,

, ,

min, min,

, ,

" "," "

" " " "

" " " "

x

s k s k

x

s k s kx x

s k s k s k s k

y y

s k s k s k s k

A D

x x DL R

x x Dx x D x x D

A D

y y D y y D

10

Disjunctions

Plant unit

Cooling

tower

At

1

At

2

HP

M.

B. Administrative

Building

Control

Room

Plant unit

Cooling

tower

At

1

At

2

HP

W.

H. Administrative

Building

Control

Room

x (m)

(85,123)

50 100

50

10

0

1st Approach – Distance based

11

2nd Approach (Overpressure based) consequence modeling

BLEVE VCE

Probability of Structure Damage

)ln(92.28.23Pr op

12

)}(exp{1 0

b

xx

ay

Sigmoid equation

a b x0

BLEVE 1.00 -8.01 64.69

VCE 1.01 -64.88 513.22

x (m)

y (

m)

(60,100)

Plant unit

Cooling

Tower

At

1

At

2

HP

W.

H.

Administrative

Building

Control

Room

50 100

50

10

02nd Approach – Overpressure based

13

Min (Land cost + ∑Interconnection costi,j + ∑PSDCi )

Potential Structure Damage Cost of i-th facility (PSDCi) =

Plant lifetime X Incident outcome frequency

X % of structural damage X Fci

Constraint: 5 meters separation between all facilities

Separation distance from the property boundary

3rd Approach – integrated with W.F. - formulations

14

Min (Land cost + ∑Interconnection costi,j + ∑PSDCWi )

Potential Structure Damage Cost of i-th facility (PSDCWi) with Weighting Factor

PSDCWi = Plant lifetime X Incident outcome frequency X % of structural damage X FCi X WFi

Objective function

i Type Population Weighting Factor, WFi

1 Control room (non-pressurized) 10 100

2 Administrative building 15 150

3 Warehouse 2.5 25

4 High pressure storage sphere 0 20

5 Atmospheric liquid storage tank 1 0 10

6 Atmospheric liquid storage tank 2 0 10

7 Cooling tower 0 1

i Type Probit function Explosion a b x0

1General

Building-23.8+2.92ln(p0)

BLEVE 1.000 -8.009 64.694

2VCE 1.006 -64.89 513.22

3

4Pressurized

vessel-42.44+4.33ln(p0)

BLEVE 1.005 -2.558 33.838

VCE 1.014 -23.22 208.78

5Atmospheric

Vessel-18.96+2.44ln(p0)

BLEVE 1.019 -10.05 66.186

6VCE 1.002 -74.53 524.12

7

3rd Approach – Probit functions

(Salzano, 2006)

15

)}(exp{1 0

b

xx

ay

3rd Approach – integrated with W.F.

16

x (m)

y (

m)

(60,100)

Plant unit

Cooling

Tower

At

1

At

2

HP

W.

H.

Administrative

Building

Control

Room

50 100

50

10

0

2. Overpressure estimation approach

x (m)

y (

m)

(85,124)

Plant unit

Cooling

Tower

At

1

At

2

HP

W.

H.

Administrative

Building

Control

Room

50 100

50

10

0

3. Integrated approach

Plant unit

Cooling

tower

At

1

At

2

HP

M.

B. Administrative

Building

Control

Room

Plant unit

Cooling

tower

At

1

At

2

HP

W.

H. Administrative

Building

Control

Room

x (m)

(85,123)

50 100

50

10

0

1. Distance-based approach

Evaluation - Asset Visualization Solution -FLACS

17

Facility #, i

Overpressure of the 1st layout result (barg)

Overpressure of the 2nd layout result (barg)

Overpressure of the 3rd layout result (barg)

Control rm 0.130 0.185 0.113Admin bld 0.092 0.131 0.089

Warehouse 0.086 0.116 0.091High press. sphere 0.218 0.241 0.215

Atm tank 1 0.233 0.207 0.226Atm tank 2 0.254 0.225 0.269

Cooling tower 0.170 0.355 0.170

Use RealityLINx 5.3 (INOVx Inc.) to obtain real geometry of the process plant

Summary Comparing three different approaches to choose the best layout

Evaluating layouts with RealityLINX and FLACS

S. Jung et. al. A New approach to Optimizing the Facility Siting and Layout for Fire and Explosion Scenarios,

I&EC Research, submitted

Conclusion

18

Compare with Grid-based plane approach

Propylene plant case study

Developing solver options

Comparing three different approaches to choose the best layout

Case study for Hexane-Heptane separation plant was demonstrated to obtain the optimal layout of 7 facilities around the process unit using MINLP

Evaluating layouts with RealiLINX and FLACS

Future work

Dr. M. Sam Mannan

Dr. Richart Vazquez

Christian Diaz Ovalle

All members of MKOPSC

Acknowledgements

Thank you!