Professional Documents
Culture Documents
CO7 Optimisation
CO7 Optimisation
FORMULATION- 4
Why OPTIMIZE?
• Monetary reason
– Largest production
– Greatest profit
– Maximum cost
– Least energy usage
– Etc
• Goal of optimization:
– To find the values of the variables in the process that yield the best value of
the performance criterion.
– Usually involved a trade of between capital and operating costs.
• Optimization
– Is concerned with selecting the best among the entire set of possible
solutions by efficient quantitative methods.
– Decision - use of past experience (engineering judgment)
Three levels optimization
• Management
– Makes decision concerning project evaluation, product selection,
corporate budget, investment in sales versus research and development
and etc.
– Information is qualitative and has a high degree of uncertainty.
• Process design and equipment specification
– Concerned with the choice of a process and nominal operating conditions.
– Resolved with the aid of process design simulators or flowsheeting
programs.
– Needs iteration to arrive at a desirable process flowsheet.
– Performed prior to the implementation of the process
• Plant operations
– Concerned with operating controls for a given unit at certain T, P, F etc
that are the best in some sense.
– Is carried out monthly, weekly, daily, hourly or even every minute at the
extreme – (real time optimisation)
– Deals with the allocation of raw materials on a daily or weekly basis
– Also concerned with the overall picture of shipping, transportation, and
distribution of products to engender minimal costs.
Optimization in engineering
Decision variables
FORMULATION- 34
DEFINITION & JARGON
Constraints
M02- 35
FEATURES OF MATHEMATICAL
PROGAMMING
min f(xi)
xi vector of decision variables to be chosen to find the min. or
max. of f(x)
equality
constraints i = 1, 2, … i i, j, k are the
# of decision
s.t. variables and
in-equality hj(xi) = 0 j = 1, 2, … j constraints
constraints
gk(xi) 0 k = 1, 2, … k
FORMULATION- 36
Optimization in engineering
Internal restrictions
- Imposed by designer
- Set by design relations
- Arise from equipment design
- Material properties (T,P)
- Positive flows
- Capacities (min, max)
- Regimes (flooding, weeping)
Equality Constraints
Equality constraints, hk (x) = 0, represent process restrictions:
For Example:
• Mole fraction flowrate must be greater than or equal to zero (no negative
flowrates are allowed)
• Pressure in a particular vessel cannot exceed an upper limit
• A reactor volume cannot exceed the maximum currently manufactured
• Production rates and product purities must all be greater than minimum
values.
OPTIMIZATION FORMULATION
subject to g1(x1,x2,……,xN)≤0
Inequality g2(x1,x2,……,xN)≤0
Constraints
gm(x1,x2,……,xN)≤0
h1(x1,x2,……,xN)=0
Equality Constraints h2(x1,x2,……,xN)=0
hE(x1,x2,……,xN)=0
FORMULATION- 41
Basic Jargons/Terminology
• Feasible solution is a set of values of the variables that satisfy all the
constraints
• Local Optimal solution provides an optimal value for the
• If other local optimal solution exist in the feasible region, one or more
could be a global optimal solution.
• Linear programming (LP) problem is when the objective function, equality
constraints, and inequality constraints are linear with respect to the
variables, x.
• Unconstrained optimisation is an optimisation problem without equality
constraints, inequality constraints and lower and upper bounds.
• Constrained optimisation is vice versa to unconstrained optimisation.
Feasible region for an optimisation problem involving
two independent variables
Nonlinear
Inequality
constraints
Linear
equality
constraints
Axis is
linear
Inequality
constraints Feasible region is along the heavy line
Axis is
linear
Inequality
constraints
Linear Objective Function
Maximum objective function is desired.
f(x) f(x)
x xopt = xU x
b) x1 + x2 2
3x1 + x2 = 3
x1, x2 0
Example 1
M02- 47
GENERAL PROCEDURE FOR
SOLVING OPTIMIZATION PROBLEM
PROBLEM
STEP 2. DEFINE THE OBJECTIVE FUNCTION
FORMULATION
LP-48
sOLUTION EXAMPLE 1:
Model Formulation
Basis: 1 kg solution
Grade 1
x1 kg Step 1: Define the decision variables
0.80 MEA, 0.20 water
1 kg MEA solution
80 cents/kg x1 Amount of grade 1 (kg)
MIXER
water content 25 wt. %
x2 Amount of grade 2 (kg)
Grade 2
x2 kg z Cost of 1 kg solution (cents)
0.68 MEA, 0.32 water
60 cents/kg
x1
Any optimal solution must lie within
the feasibility region!
SOLUTION EXAMPLE 1: GRAPHICAL
SOLUTION
x2
Feasibility region
0.2
0x
1+
0.3
is this heavy line
2x
2=
0.2
5
Optimal Point
Intersection between
x1 + x2 = 1
and
0.20x1 + 0.32x2 = 0.25
In addition
70 < zmin < 75
SOLUTION EXAMPLE 1: GRAPHICAL
SOLUTION
x1
x2
+x
2=
1
0.2
0 x1 z* = 71.6
+0
.32
x 2=
0.2
Optimal
5
point
x2* = 0.42
x1
x1* = 0.58
EXAMPLE 2
FORMULATION- 54
GENERAL PROCEDURE FOR
SOLVING OPTIMIZATION PROBLEM
objective function 10 HB + 18 FF
FORMULATION- 56
Example 3
FORMULATION- 57
Solution Example 3
STEP 1: Define the desicion variables
x1 = bags of Super-gro
x2 = bags of Crop quick
• Inequality constraint
2x1 + 4x2 16 (nitrogen constraint)
4x1 + 3x2 24 (phosphate constraint)
FORMULATION- 58
Solution Example 3
FORMULATION- 59
SOLUTION EXAMPLE 3:
GRAPHICAL SOLUTION
STEP 5: Apply suitable optimization technique
4x1 + 3 x2 =24
2x1 + 4x2 = 16
STEP 6: Perform sensitivity analysis e.g. What if the cost of the raw material
(e.g. super-grow) is increased by 10%
FORMULATION- 60
SOLUTION TERMINOLOGY
1 1 0 4
2 0 2 12
3 3 2 18
Profit per 3 5 LP-62
MODEL FORMULATION
LP-63
GRAPHICAL SOLUTION METHOD
1. Plot each of the constraints
2. Identify the feasible region
3. Plot isoprofit line (objective function) to find the point on boundary of
this space that maximizes (or minimizes) value of objective function
4. Align a straight edge with the plotted objective function, move it towards
the inside of the feasible region
5. Move in the direction that will increase the value of the objective
function (if maximizing) and move in the direction that will decrease the
value of the objective function (if minimizing)
6. Move the straight edge until it gets out of the feasible region. The
last point that the straight edge touches before leaving the feasible region is
the optimal solution. If the straight edge touches a whole line segment
before it leaves the feasible region, then any point on that line is
optimal. In that case we have alternative optima.
LP-64
FEASIBLE SOLUTION
A SOLUTION THAT SATISFIES THE CONTRAINTS
x2
3x1 + 5x2 = 36
Feasible region
6
x1
0 1 2 3 4 5 7 8 9 10
3x1 + 5x2 = 10
LP-65
INFEASIBLE SOLUTION
A SOLUTION THAT VIOLATES ONE OF THE
CONSTRAINTS
What if there is another constraint?
x2
another constraint
6
x1
0 1 2 3 4 5 7 8 9 10
LP-66
UNBOUNDED FEASIBLE
REGION
NO SOLUTION
x2
6
x1
0 1 2 3 4 5 7 8 9 10
LP-67
UNBOUNDED FEASIBLE REGION
NO SOLUTION
x2
6
x1
0 1 2 3 4 5 7 8 9 10
LP-68
SOLUTION TERMINOLOGY
LP-69
SOLUTION TERMINOLOGY
x2
Optimal Solution
Isoprofit Line
(2,6) Corner-point feasible solution (CPF)
Vertex Extreme Point
Feasible Region
6
x1
0 1 2 3 4 5 7 8 9 10
LP-70
WORKING SESSION 2:
LP PROBLEM
RESOURCE REQUIREMENTS
Labor Clay Revenue
PRODUCT (hr/unit) (lb/unit) ($/unit)
Bowl 1 4 40
Mug 2 3 50
There are 40 hours of labor and 120 pounds of clay available each
day
Decision variables
x1 = number of bowls to produce
x2 = number of mugs to produce
Maximize Z = $40 x1 + 50 x2
Subject to
x1 + 2x2 40 hr (labor constraint)
4x1 + 3x2 120 lb (clay constraint)
x1 , x2 0
40 –
4 x1 + 3 x2 120 lb
30 –
Optimal value
X1 = 24 bowls Z = $50(24) + $50(8) =
20 – X2 = 8 mugs $1,360
10 – x1 + 2 x2 40 hr
| | | | | |
0–
10 20 30 40 50 60 x1
x1 = 0 bowls
x2 x2 =20 mugs
x1 = 224 bowls
Z = $1,000
x2 =8 mugs
40 –
x1 = 30 bowls
Z = $1,360
30 – x2 =0 mugs
20 – A Z = $1,200
10 – B
| | | C|
0–
10 20 30 40 x1
Two Crude Petroleum runs a small refinery on the Texas coast. The refinery
distills crude petroleum from two sources, Saudi Arabia and Venezuela, into
three main products: gasoline, jet fuel and lubricants.
The two crudes differ in chemical composition and thus yield different product
mixes. Each barrel of Saudi crude yields 0.3 barrel of gasoline, 0.4 barrel of
fuel jet and 0.2 barrel of lubricants. On the other hand, each barrel of
Venezuela crude yields 0.4 of gasoline, 0.2 barrel of fuel jet and 0.3 barrel of
lubricants. The remaining 10% of each barrel is lost to refining.
The crudes also differ in cost and availability. Two Crude can purchase up two
9000 barrels per day from Saudi Arabia at $20 per barrel. Up to 6000 barrels
per day of Venezuela petroleum can are also available at the lower cost of
$15 per barrel because of the shorter transportation distance.
x1 ≤9 (Saudi availability)
x2 ≤6 (Venezuelan availability)
x1: barrels x
of1,Saudi
x2 crude refined
≤ 0per day (in thousands)
(non-negativity)
x2: barrels of Venezuelan crude refined per day (in thousands)
X2 ≤ 6
The optimal solution is
x1
+0
Optimal solution
only feasible solution
≥1
x1 = 2
.5
x2 = 3.5
achieving the optimal
value
X1 ≤ 9
0.3
0.2 x
x 1 +0
1 +0 .4
.3 x 20x1 + 15x2 = 220
x2 2 ≥2
≥0
.5 20x1 + 15x2 = 180
6
x1
0 1 2 3 4 5 7 8 9 10
20x1 + 15x2 = 92.5 20x1 + 15x2 = 100
LP-77
Alternative Optimal Model
Suppose that the crude price from Venezuela drop to $10 per barrels,
the new objective function is,
x1 ≤9 (Saudi availability)
x2 ≤6 (Venezuelan availability)
x1, of
x1: barrels x2Saudi crude ≤refined
0 per day (non-negativity)
(in thousands)
x2: barrels of Venezuelan crude refined per day (in thousands)
x2
X2 ≤ 6
If the optimal-value
contour intersects at
alternative optimal solution
more than one point, the
model has alternative
optimal solutions.
0.4
x1
X1 ≤ 9
+0
.2 x 2
0.3
0.2 x
≥1
x 1 +0
+0
.5
6
x1
0 1 2 3 4 5 7 8 9 10
20x1 + 10x2 = 75
LP-79
Infeasible Model
Consider modifying our Two Crude case so that only 2000 barrels per
day are available from each source. The resulting model is:
x1 ≤2 (Saudi availability)
x2 ≤2 (Venezuelan availability)
x1, of
x1: barrels x2Saudi crude ≤refined
0 per day (non-negativity)
(in thousands)
x2: barrels of Venezuelan crude refined per day (in thousands)
new constraint
LP-80
Infeasible SOLUTION
x2
X1 ≤ 2 An optimization model is
0.4
x1
infeasible if no choice of
+0
No feasible region!
.2 x 2
decision variable
≥1
0.3 X2 ≤ 2
x
0.2 1 +0
x1 .4
+0 x
.3
2 ≥2
x
2 ≥0
.5
x1
0 1 2 3 4 5 6 7 8 9 10
LP-81
Unbounded Model
x2 X1 is infinite
0.4
X2 ≤ 6
x1 +0
.2 x 2
70
≥1
60
.5
50
40
30
0.3
0.2 x 20
x1 1 +0
+0 .4
.3 x 10
x 2 ≥2
2 ≥0
.5 0
6
-10
x1
0 1 2 3 4 5 7 8 9 10
During the 2002 Winter Olympics in Salt Lake City, Utah, a local plant
X received a rush order for 100 gals of A containing 4.0% vol%
alcohol. Although no 4% A was in stock, large quantities of A-4.5 with
4.5% alcohol at a price of $6.40/gal and A-3.7 with 3.7% alcohol
priced at $5.00/gal were available, as well as water suitable for
adding to the blend at no cost. The plant manager wanted to use at
least 10 gal of A-4.5. Neglecting any volume change due to mixing,
determine the gallons each of A 4.5, A-3.7, and water that should be
blended together to produce the desired order at the minimum cost.
Use 6 steps approach to solve optimization problem.
Step 4 : Simplification.
The problem can be reduced to two decision variables by solving eq 3 for
VA-3.7,
VA-3.7 = 100 – VA-4.5 – VW (eq 4)
And substituting it into eq. 1 and 2 to give the following restatement of this
problem:
The optimal volume of A-3.7 need only to be calculated from eq 7, after the
optimal volumes of A-4.5 and water have been determined. Since the
objective function, the equality constraints, and the lower and upper bound are
all linear, this constitutes an LP problem.
Solution to Example 5 (Cont.)
Feasible region
80
V A4.5 constraint
V, gallons
60 VA4.5
VW
40 VA3.7
20
V A4.5 constraint
V A3.7 & V W 0
constraint
540 545 550 555 560 565 570 575
-20
Cost, $
With the constrained given, the optimal solution is when V A4.5 is 37.5 gal,
VA3.7 is 62.5 and VW is 0 with the minimum cost at $552.50.
Example 6
A chemical plant makes three products
(E, F, G) and utilizes three raw Maximum
materials (A, B, C) in limited supply. Raw available, Cost,
Each of the three products is produced Material lb/day cent/lb
in a separate process (1, 2, 3); a
schematic of the plant is shown in A 40,000 1.5
Fig.2. The available materials A, B and
B 30,000 2.0
C do not have to be totally consumed.
Find the optimum production to C 25,000 2.5
maximize the total profit per day in $/day
Reactant
requirement (lb) Processing Selling price
Process Product per lb product cost of product
1 E 2/3 A, 1/3 B 1.5 cent/lb E 4.0 cent/lb E
2 F 2/3 A, 1/3 B 0.5 cent/lb F 3.3 cent/lb F
3 G 1/2 A, 1/6 B, 1/3 C 1.0 cent/lb G 3.8 cent/lb G
EXAMPLE 6 (CON’T)
Process 1: A + B E
The reactions involving Process 2: A + B F
A, B and C Process 3: 3A + 2B + C G
are as follows:
A
1 E
2 F
B
3 G
C
Figure 2
Example 6(Cont)
Step 1 : Define the decision variable, XA, XB, Xc, XE, XF, XG
Step 2 : Define the objective function
Profit = Income – raw material cost – Processing cost
• Income = 0.04 xE + 0.033 xF + 0.038 xG
• Operating costs in $ per day include:
– (a) Raw material costs : 0.015xA + 0.02xB + 0.025xC
– (b) Processing costs: 0.015xE + 0.005xF + 0.01xG
Therefore
Max f(x) = (0.04xE + 0.033xF + 0.038xG) – (0.015xA + 0.02xB + 0.025xC + 0.015xE
+ 0.005xF + 0.01xG)
Step 4: Simplification
Not applicable