You are on page 1of 22

De NoVo programming

Ha Thi Xuan Chi, PhD


MILAN ZELENY

Fordham University, New York, USA


De NoVo programming (Zenleny -1981, 1986)
New concepts of optimality, new ways of resolving MCDM conflicts,
and conditions for optimal and continuous system improvement:
 Optimizing a given system
 Designing an optimal system
Optimality
1. A given system:
a) Single objective function: Problem of measurement and search
(exemplified by linear programming ).
b) Multiple objective function: problem of decision making and
search (MCDM methodologies)
Optimality (cont.)
2. An optimal system
a) Single objective function: problem of design and measurement
( De Novo programming)
b) Multiple objective function: problem of design and decision
making ( De Novo programming)
De Novo Programming procedure

Step 1: Formulate a linear programming problem:

max 𝑧=𝐶𝑥

where
De Novo Programming procedure
Step 2: Find the aspiration level vector z* by solving each objective
function of a system separately as

max𝑧=𝐶𝑥
z*=(z1*,.., zq*) be the q- objective value for the ideal system
with respect to B.

𝑤h𝑒𝑟𝑒
Multiple Criteria De Novo Programming procedure (cont.)

Step 3: Identify the minimum budget B*and its corresponding resource allocation
with the aspiration level

𝑀𝑖𝑛 B ∗ = 𝑣𝑥
𝑞
z=(z1 ,. ., zq)∈R
Multiple Criteria De Novo Programming procedure (cont.)

Step 4: Use the optimum –path ration to obtain the final solution:

x=r*.x*
b=r*.b*
z=r*.z*

where

r∗=B/B*
Example of optimal design*
Table 1. Current portfolio of available capacities

Machine type Available Unit price($100 per hour)


Milling machine 1400 0.75
Lathe 1000 0.6
Grinder 1750 0.35
Jig saw 1325 0.5
Drill press 900 1.15
Band saw 1075 0.65

B=1400*0.75+1000*0.6+1750*0.35+1325*0.5+900*1.15+1075*0.65
=4658.75

( *adapted from “Optimal system design with multiple criteria: De Novo programming approach”)
Example of optimal design (cont.)
Table 2. Technological coefficients

Machine type Product 1 Product 2 Product 3


Milling machine 12 17 0
Lathe 3 9 8
Grinder 10 13 15
Jig saw 6 0 16
Drill press 0 12 7
Band saw 9.5 9.5 4
Example of optimal design (cont.)
Formulate the linear programming problem

𝑀𝑎𝑥𝑖𝑚𝑖𝑧𝑒
𝑆𝑢𝑏𝑗𝑒𝑐𝑡𝑡𝑜 :
Example of optimal design (cont.)
Reconstruct the linear programming problem

𝑀𝑎𝑥𝑖𝑚𝑖𝑧𝑒 𝑀𝑎𝑥𝑖𝑚𝑖𝑧𝑒
𝑆𝑢𝑏𝑗𝑒𝑐𝑡𝑡𝑜 : 𝑆𝑢𝑏𝑗𝑒𝑐𝑡𝑡𝑜
𝑀𝑎𝑥𝑖𝑚𝑖𝑧𝑒
Subject to:

≤ 4658.7
Max Z1=

Subject to:

≤ 4658.7

Solution: x1=0, x2=109.167, x3=0

Z1*=10916.696
Max Z2=

Subject to:

≤ 4658.7

Solution:
x1=198.4537,x2=0,x3=0

Max Z2=18257.738
Max Z3=

Subject to:

≤ 4658.7

Solution:
x1=0, x2=0, x3=162.324
Z3=12174.3031
Example of optimal design (cont.)
• Construct a metaoptimum problem

𝑀𝑖𝑛

𝑠𝑢𝑏𝑗𝑒𝑐𝑡 𝑡𝑜
Example of optimal design (cont.)
• Solution:

∗ ∗ ∗
𝑥 =131.341,𝑥 =29.683,𝑥 =78.976
1 2 3

[ ][ ] [ ][

𝑏1
∗ 2080.703
𝑏

]
2 1292.978 ∗
𝑏

3
=
2883.929 𝑧 1 10916.813

𝑏 2051.662 ∗
𝑏
4

5
909.028 𝑧 2
= 18257.933
1845.632
𝑏

6
𝑧
∗ 12174.433
3
Example of optimal design (cont.)
• r=B/B*=4658.75/6616.319 =0.70385

[] [ [] ] [ ] [ ] [ ]
∗ ∗
𝑥1 𝑥 92.48
1 𝑧1 𝑧1 7686


𝑥2 =𝑟 × 𝑥 = 20.90 𝑧 2 =𝑟 × 𝑧 2 = 12855.89
2
55.61
∗ 𝑧3 𝑧3
∗ 8572.4
𝑥3 𝑥 3

[ ] [ ][ ]

Milling
𝑏1
𝑏1
1465.06
machine 1400

𝑏2 𝑏2 910.42 Lathe 1000

𝑏3 𝑏3 2030.65 Grinder 1750
=𝑟 × ∗
=
𝑏4 𝑏4
1444.64 Jig saw 1325
𝑏5 𝑏
∗ 604.07
𝑏6
5
∗ ¿ 1 2 99. 55
Drill press 900
𝑏6 Band saw 1075
Example of optimal design (cont.)
Table 4 : Metaoptimal design and its costs

Machine type Recommended Cost ($)


availability (machine-h/w)
Milling machine 2080.5963 1560.527
Lathe 1292.925 775.7868
Grinder 2883.8523 1009.375
Jig saw 2051.6661 1025.831
Drill press 908.95709 1045.382
Band saw 1845.5731 1199.661
Total 11063.57 6616.563
Ending slide
Put your last thoughts or wrap up here

You might also like