MULTI-OBJECTIVE MIXED INTEGER

PROGRAMMING APPROACH FOR

FACILITY LAYOUT DESIGN BY
CONSIDERING CLOSENESS
RATINGS, MATERIAL HANDLING,
AND RELAYOUT COST
Yoga Satrio Wiwoho – 2015
Supervisor: Muhammad Ridwan Andi Purnomo
December 7th, 2015
International Program of Industrial Engineering Department
Faculty of Industrial Technology
Universitas Islam Indonesia
 Intro:

Overview Conclusion & Further Research

• Background
• Research Objective
• Related Works
• Research Location &
Objects
• Problem Definition

Discussion: Mathematical
Model &
• Objective Value Comparison Methodolgy

Case study:
• Initial Layout
• Departments
• Assumptions
• New Layout
Background

Production system Modelling

• Saving time & cost for seeing the new
layout?
• Facility Layout

Mixed Integer • Getting the optimal layout?

Programming

Multi-objective • Look for a more

model practical model?
Research Is there an improvement in the new layout by using Mixed
Integer Programming approach?
Questions If yes, how is the improvement?
 Scope of • Model 
– Multi row
Problem – Single floor

Definition – Unequal rectangle area

• Assignment problem, where a set of n facilities will be assigned

to m locations.

• The cost of relayout objective will focus on the cost of

moving machine and production loss as the detail will be
explained later.
Related Works
Singh & Sarma
(2006)

Konak et al.
Sherali et al. (2005)  for
FBS layout Jain, Khare, &
(2003) Mishra (2013)
Meller (1999)

Montreuil
(1991) Multi-objective
as future
research
Heragu &
Kusiak
(1991) :
M1, M2, &
M3 MIP development Literature Reviews
 State of the Art
Research proposed
Sha & Chen (2001) Chen & Sha (2005) Shah et al. (2011) Aiello et al. (2012) Lenin et al. (2013)
(2015)
Multi objective V V V V V V
facility layout
problems
Material Handling V V V V
Cost objective
Closeness/Adjacency V V V
Rating objective
Probability of V
Superiority objective
Aspect ratio objective V
Distance Request V
objective
Total flow distance V
objective

MIP developmentV
Total number of
machines objective
Total Investment cost V
objective
Total Re-Layout V
objective
 Research • Location: UD Rekayasa Wangdi (Cambahan, Nogotirto,
Location and
Gamping, Sleman, DIY). Rekayasa Wangdi is a company focusing in
producing machines for industries and personal usage related in
manufacturing, trade, and service of food and beverage.
Objects • The layout for the production of concrete brick dough
stirring machine  company’s best seller
Mathematical w

𝑐𝑖𝑗
= weighting score of each objective

= cost of material handling from facilities i to j

Model 𝑓𝑖𝑗

𝑃𝑙𝑖
= frequency of material flow from facilities i to j

= production loss from moving facilities i

(Parameter) 𝑡𝑖𝑗

𝑐𝑚1𝑖𝑗
= time spent for moving facilities i to j

= cost of moving facilities i to j (where it is varied to distance)

𝑐𝑚2𝑖𝑗 = fixed cost of moving facilities i to j (where it is not influenced by distance)

𝑟𝑖𝑗 = adjacency score between facilities i to j

𝑎𝑖 = area of department i, where 𝑎𝑖 > 0

∝𝑖 = maximum permissible ratio between the longest and shortest side of department i. i.e.,

max 𝑙𝑖𝑠 / min 𝑙𝑖𝑠 ≤ ∝𝑖

𝑠 𝑠

𝑙𝑏𝑖 = lower bound of department’s length

𝑢𝑏𝑖 = upper bound of department’s length

𝐻𝑠 = maximum permissible total length (building criteria)

s = x, y coordinate
Mathematical 𝑙𝑖𝑥

𝑦
= half-length (correspond to x ordinate)

Model
𝑙𝑖 = half-width (Correspond to y ordinate)

𝑐𝑖𝑥 = centroid of department i in x ordinate

(Variables)
𝑦 = centroid of department i in y ordinate
𝑐𝑖

𝑠 = distance between facilities i and j

𝑑𝑖𝑗

𝑠
𝑧𝑖𝑗
1, 𝑖𝑓 𝑑𝑒𝑝𝑎𝑟𝑡𝑚𝑒𝑛𝑡 𝑖 𝑖𝑠 𝑓𝑜𝑟𝑐𝑒𝑑 𝑡𝑜 𝑝𝑟𝑒𝑐𝑒𝑑 𝑑𝑒𝑝𝑎𝑟𝑡𝑚𝑒𝑛𝑡 𝑗 𝑖𝑛
𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 𝑠

0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
Mathematical min 𝑍 = 𝑤1 𝑂1 − 𝑤2 𝑂2 + 𝑤3 𝑂3𝑎 + 𝑤3 𝑂3𝑏 (01)

Model 𝑛−1 𝑛 Material handling cost

(Objective)
𝑥 𝑦
𝑂1 = ෍ ෍ 𝑐𝑖𝑗 𝑓𝑖𝑗 (𝑑𝑖𝑗 + 𝑑𝑖𝑗 )
minimization
𝑖=1 𝑗=1+1

𝑛−1 𝑛
𝑥 𝑦 Adjacency score maximization
𝑂2 = ෍ ෍ 𝑟𝑖𝑗 (𝑑𝑖𝑗 + 𝑑𝑖𝑗 )
𝑖=1 𝑗=1+1

𝑛−1 𝑛 Minimization of moving machine

𝑥 𝑦
𝑂3𝑎 = ෍ ෍ 𝑐𝑚1𝑖 (𝑑𝑖𝑗 + 𝑑𝑖𝑗 )
cost (influenced by distance)
𝑖=1 𝑗=1+1

Minimization of moving machine

𝑛 cost (not influenced by distance)

𝑂3𝑏 = ෍ 𝑐𝑚2𝑖 + (𝑝𝑙𝑖 𝑡𝑖 )
and production loss of moving
𝑖=1

machine
Mathematical • Area constraint : Eq. (06) and (07)
• Departments i and j are separated : Eq. (08) and (09)

Model • Departments overlapping prevention : Eq. (10)

• Linearization of distance’s absolute value from the centroid i and j

(Constraints) difference : Eq. (11) and (12)

• Departments is inside building : Eq. (13)
• Length of departments not exceeding boundary set : Eq. (14)
• Value of lower and upper bound of the department’s length : Eq. (15) and (16)
• Constraints of the distance between departments : Eq. (17)
• Centroid is not a negative value : Eq. (18)
• Decision variable to be binary : Eq. (19)
• p and q strategy constraints : Eq. (20) – (22)
• Valid inequalities : Eq. (23) and (24)
Mathematical
𝑦
𝑎𝑖 𝑙𝑖𝑥 + 4𝑥ҧ 2 𝑙𝑖 ≥ 2𝑎𝑖 𝑥ҧ ∀ 𝑙𝑏𝑖𝑥 ≤ 𝑥ҧ ≤ 𝑢𝑏𝑖𝑥 , i (06)

𝜆
𝑥ҧ = 𝑙𝑏𝑖𝑥 + 𝑢𝑏𝑖𝑥 − 𝑙𝑏𝑖𝑥 ∀ 𝜆 = 0, 1, … , Δ − 1,

Model
Δ−1 (07)

𝑓𝑜𝑟 𝑎𝑛𝑦 𝑠𝑒𝑙𝑒𝑐𝑡𝑒𝑑 𝑖𝑛𝑡𝑒𝑔𝑒𝑟∆≥ 2

(Constraints)
𝑦
𝑠 (08)
෍ 𝑧𝑖𝑗 + 𝑧𝑗𝑖𝑠 ≥ 1 ∀ 𝑖<𝑗
𝑠=𝑥

𝑠
𝑧𝑖𝑗 + 𝑧𝑗𝑖𝑠 ≤ 1 ∀ 𝑗 > 𝑖, 𝑠 (09)

𝑐𝑖𝑠 + 𝑙𝑖𝑠 ≤ 𝑐𝑗𝑠 − 𝑙𝑗𝑠 + 𝐻 𝑠 1 − 𝑧𝑖𝑗

𝑠
∀ 𝑖 ≠ 𝑗, 𝑠 (10)

𝑠
𝑑𝑖𝑗 ≥ 𝑐𝑖𝑠 − 𝑐𝑗𝑠 ∀ 𝑓𝑖𝑗 ≠ 0, 𝑗 > 𝑖, 𝑠 (11)

𝑠
𝑑𝑖𝑗 ≥ 𝑐𝑗𝑠 − 𝑐𝑖𝑠 ∀ 𝑓𝑖𝑗 ≠ 0, 𝑗 > 𝑖, 𝑠 (12)

𝑙𝑖𝑠 ≤ 𝑐𝑖𝑠 ≤ 𝐻 𝑠 − 𝑙𝑖𝑠 ∀ 𝑖, 𝑠 (13)

𝑙𝑏𝑖 ≤ 𝑙𝑖𝑠 ≤ 𝑚𝑖𝑛 𝑢𝑏𝑖 , 𝐻 𝑠 /2 ∀ 𝑖, 𝑠 (14)

𝑢𝑏𝑖 = 𝑚𝑖𝑛 ∝𝑖 𝑎𝑖 , 𝑚𝑎𝑥𝑠 𝐻 𝑠 /2 ∀ 𝑖, 𝑠 (15)

𝑙𝑏𝑖 = 𝑎𝑖 /(4𝑢𝑏𝑖 ) ∀ 𝑖 (16)

𝑠
0 ≤ 𝑑𝑖𝑗 ≤ 𝐻 𝑠 − (𝑙𝑏𝑖 +𝑙𝑏𝑗 ) ∀ 𝑖 ≠ 𝑗, 𝑠 (17)

𝑐𝑖𝑠 ≥ 0 (18)

𝑠
𝑧𝑖𝑗 ∈ 0,1 ∀ 𝑖 ≠ 𝑗, 𝑠 (19)
Mathematical 𝑐𝑝𝑠 ≤ 𝑐𝑞𝑠 ∀ 𝑠 = 𝑥, 𝑦
(20)

Model
𝑥 𝑦 (21)
𝑧𝑞𝑝 = 𝑧𝑞𝑝 = 0

σ𝑦𝑠=𝑥 𝑐𝑞𝑠 − 𝑐𝑝𝑠 ≥ 𝑚𝑖𝑛 𝑙𝑏𝑝𝑥 + 𝑙𝑏𝑞𝑥 , 𝑙𝑏𝑝𝑦 + 𝑙𝑏𝑞𝑦 (22)

(Constraints) 𝑠
V2: 𝑑𝑖𝑗
𝑠
B2: 𝑑𝑖𝑗 ≥ (𝑙𝑏𝑖𝑠 +𝑙𝑏𝑗𝑠 ) ∀ 𝑓𝑖𝑗 ≠ 0, 𝑗 > 𝑖, 𝑠

≥ (𝑙𝑖𝑠 +𝑙𝑗𝑠 ) − 𝑚𝑖𝑛 𝑢𝑏𝑖𝑠 + 𝑢𝑏𝑗𝑠 , 𝐻 𝑠 (1 − 𝑧𝑖𝑗

𝑠
− 𝑧𝑗𝑖𝑠 )
(23)

(24)

∀ 𝑓𝑖𝑗 ≠ 0, 𝑗 > 𝑖, 𝑠
Data Collection Primary data:
– Number of facilities
– Frequency of material flow
– Area of each machine/facility with its length and wide
– Adjacency rating between facilities
– Cost of moving machine/facility
– Production capacity for each machine/facility
– Time of moving facilities
– Allowed area of the building for production facilities/machines

Secondary data:
– Bill of Material of the product produced
– Flow diagram of the material produced
– Operation Process Chart (OPC) of the product
Data Collecting • Observation
Direct observation is conducted to the production floor so the layout data
Method and any other data, especially the quantitative one, can be collected.

• Interview
Interview is conducted toward the owner/manager and the operator of the
production itself, so the data that cannot be observed by own can be
gathered.
Initial Layout
Departments
Involved
𝒚
No Dept. L (m) H (m) 𝒄𝒙𝒊 𝒄𝒊 Cost of moving machine (IDR/meter)
1 Cutting machine 12 8 16,55 5,725 7.200.000 (fixed)
2 Cutting dept. 3 4 15,2 6,05 Equal to material handling cost/meter
3 Drill 3 2,5 15,1 6,375 Equal to material handling cost/meter
4 Welding 8 2,5 17,05 6,275 Equal to material handling cost/meter
5 Circle 9 4,5 16,4 6,825 Equal to material handling cost/meter
6 Grinding 6 2,5 15,65 6,375 Equal to material handling cost/meter
7 Bender 3 2,5 15,825 6,625 Equal to material handling cost/meter * 2
8 Roll plate 4 2 16,85 6,5 7.200.000 (fixed)
9 Painting 6 6 15,65 5,95 Equal to material handling cost/meter
10 Bending 7 4,5 16,3 6,375 7.200.000 (fixed)
 Departments
Involved
• Production loss per minute for each departments • Moving time of machine:

Operation Production Operation time/total Dept. Moving Time

Dept.
Time (min) loss/min (%) (min)
1 35 0,010 1% 1 1.920
2 103 0,030 3% 2 5
3 392 0,112 11% 3 5
4 1.640 0,470 47% 4 3
5 529 0,152 15%
5 5
6 198 0,057 6%
6 3
7 290 0,083 8%
7 15
8 30 0,009 1%
8 1.920
9 240 0,069 7%
10 30 0,009 1% 9 7
Sum 3.487 1 100% 10 1.920
 Inter-
departments
• Flow frequency between departments: • Closeness ratings between departments:
Dept. 1 2 3 4 5 6 7 8 9 10 Dept. 1 2 3 4 5 6 7 8 9 10
1 0 2 1 1 0 4 3 2 2 3 2 4 1 3
2 0 3 3 3 2 4 0 4 2 2 3 2 2 1 2
3 1 0 10 1 1 3 3 4 0 3 3 5 2 2 1 3
4 1 0 1 4 2 2 3 0 2 4 2 2 1 2
5 2 6 0 8 4 5 2 2 3 2 0 3 3 3 1 2
6 5 7 0 1 6 3 3 5 4 3 0 2 2 1 5

7 7 0 7 2 2 2 2 3 2 0 2 1 2
8 4 2 2 2 3 2 2 0 2 2
8 1 0
9 1 1 1 1 1 1 1 2 0 3
9 0
10 1 0 10 3 2 3 2 2 5 2 2 3 0
Assumptions • Material handling cost = 1
• Value of ∆ in area constraint is 20
• The α ratio uses the initial length and widht
comparisons
• The departments are located along a maximum
𝑦
permissible length 𝐻 𝑥 = 24 and 𝐻 =18 (in meter)
• The weighing score:
– 𝑤1 = 0,4
– 𝑤2 = 0,5
– 𝑤3 = 0,1

• p and q departments for p-q strategy area the pair of

departments 3-4
 New Layout
• Variables (centroid & half-length) yielded from calculation: • The new layout:

𝒚 𝒚
Dept. 𝒄𝒙𝒊 𝒄𝒊 𝒍𝒙𝒊 𝒍𝒊
1 9,305733 10,48089 5,356049 4,480891
2 1,974842 16,48089 1,974842 1,519109
3 4,61782 2,25 1,5 1,25
4 18,66178 4,75 4 1,25
5 19,16178 8,25 4,5 2,25
6 17,66178 16,75 3 1,25
7 7,411782 4,75 1,25 1,25
8 22,85734 1,75 1,142659 1,75
9 11,66178 3 3 3
10 18,21143 13 3,549648 2,230891
Objective Value Objective Value comparison between initial and new layout:

Comparison Objective Value Difference

Initial Layout 2.159.221,50
333,72
Proposed Layout 2.158.887,79

Proposed Layout has lower score with 333,72 difference

 Conclusion &
Further
Research
Future Research:
MIP is able to optimize a multi-objective facility • Modifying the minimization of relayout cost objective
layout problem with 333,72 objective value with a dynamic machine moving time, fixed
departments, etc.
improvement
• Implementing weighting methods (ANP, AHP, MCDM,
Pareto Optimality, etc.)
THANK YOU