Optimization 240122 191010
Optimization 240122 191010
Optimization principles have historical roots, demonstrated by the ancient Roman bathhouse
inscription. Optimization is pervasive in science, engineering, and business, addressing diverse
objectives like maximum profit, minimum cost, and resource efficiency. In chemical engineering,
optimization deals with finding the best values for variables in a process to achieve a desired
performance criterion, often involving a trade-off between capitals and operating costs.
Chemical engineering problems often have multiple solutions. Optimization employs efficient
quantitative methods, with computers playing a crucial role in making computations feasible and cost-
effective. Critical analysis, insight into performance objectives, and engineering judgment are necessary
for effective optimization.
2. Why Optimize?
2.1 Engineers' Interest in Optimization
Engineers pursue optimization to enhance equipment design, improve operational efficiency, and
achieve objectives such as increased production, greater profit, and reduced energy usage. Monetary
value is a convenient measure, but optimization benefits extend beyond monetary frameworks.
Improved plant performance through optimization leads to benefits like enhanced product yields,
reduced energy consumption, increased processing rates, and longer operational durations.
Additionally, optimization contributes to reduced maintenance costs, less equipment wear, and better
staff utilization. Identifying objectives, constraints, and degrees of freedom systematically improves
design quality and decision-making speed.
Optimization benefits should be carefully predicted due to the coupling of design and operating
variables. Uncertainty in mathematical representations and dynamic plant operating conditions requires
engineers to exercise judgment. Sensitivity analysis helps evaluate the impact of uncertain parameters
on optimal design, acknowledging that optimization results provide upper limits on expectations.
3. Scope and Hierarchy of Optimization
3.1 Levels of Optimization
Optimization can occur at various levels within a company, ranging from the entire organization to
individual pieces of equipment. The scope of optimization problems can encompass the entire company,
a plant, a process, a single unit operation, or even smaller entities. The complexity of analysis depends
on the desired results, data availability, and time constraints.
Optimization at different levels operates on distinct time scales. Management decisions occur far in
advance, process design precedes implementation, while plant operations involve real-time
adjustments. The frequency of optimization varies, ranging from monthly to every minute, depending on
the operational context.
Sales Limited by Production: Incremental sales beyond current capacity offer economic
justification for design modifications.
Sales Limited by Market: Efficiency improvements become the focus when additional products
cannot be sold.
Large Unit Throughputs: High production volumes magnify small savings in production costs per
unit.
High Raw Material or Energy Consumption: Significant savings can be achieved by reducing the
consumption of items with high unit costs.
Product Quality Exceeds Specifications: Operating close to customer specifications can yield cost
savings.
Losses of Valuable Components through Waste Streams: Optimizing air-fuel ratios can minimize
material losses in waste streams.
High Labor Costs: Improved facility layouts and handling practices can reduce labor costs.
Valuable data sources for identifying optimization opportunities include profit and loss statements and
periodic operating records. Profit and loss statements contain information on sales, prices,
manufacturing costs, and profits, while operating records present material and energy balances, unit
efficiencies, and production levels.
Due to the complexity of chemical plants, complete optimization can be challenging. Sub optimization,
focusing on one phase while ignoring certain factors, becomes a practical approach. Economic and
practical constraints, time limitations, and the difficulty of obtaining quick answers often necessitate sub
optimization. While it provides a rational technique, sub optimization may not guarantee an overall
optimum for the entire system due to potential incompatibility between subsystem and overall
objectives.
Optimization finds extensive application in chemical processes and plants, contributing to various
projects. Some notable examples include:
Plant Location: Determining the optimal sites for constructing chemical plants.
Tanker Routing: Optimizing routes for the distribution of crude and refined products using
tankers.
Pipeline Sizing and Layout: Designing the size and layout of pipelines for efficient transport.
Equipment and Plant Design: Applying optimization in designing individual equipment and entire
plants.
Maintenance Scheduling: Optimizing schedules for maintenance and equipment replacement to
maximize operational efficiency.
Operational Optimization: Efficiently operating equipment such as tubular reactors, columns,
and absorbers.
Process Modeling: Evaluating plant data to construct accurate models of chemical processes.
Inventory Management: Minimizing inventory charges through effective optimization strategies.
Resource Allocation: Allocating resources or services among different processes for optimal
utilization.
Construction Planning: Planning and scheduling construction activities for efficiency.
This section provides practical illustrations of optimization in process operations and design. These
examples serve to highlight the essential features of optimization problems, offering insight into the
diverse variables, objective functions, and constraints encountered in subsequent chapters.
Optimizing insulation thickness is crucial for cost savings, particularly when fuel costs are high. The goal
is to balance the expense of insulation material with the savings from reduced heat losses. In this
example, a vessel's bare surface at 700°F experiences a heat loss of 4000 Btu/(h)(ft2) at an ambient
temperature of 70°F. Adding 1 inch of calcium silicate insulation reduces the loss to 250 Btu/(h)(ft2).
Optimization Considerations:
1. Cost Analysis:
Installed cost of insulation: $4.00/ft2
Cost of energy: $5.00/10^6 Btu
Savings per year per square foot: $164 (8760 hours of operation)
2. Payback Period:
Simplified payback calculation: $4.00/(ft2) / $164/(ft2/year) = 0.0244 years or 9 days
3. Optimization Process:
Additional insulation inches added in 0.5-in. increments.
Energy loss cost represented as a continuous curve based on heat transfer principles.
Capital costs displayed as prorated dots due to insulation longevity.
Outcome:
Optimizing the operation of a boiler is crucial in refineries and process plants due to the significant
energy consumption in utilities and powerhouse operations. Engineers face the challenge of balancing
the optimization of air-fuel ratio for thermal efficiency with
the need to comply with environmental regulations,
particularly in reducing nitrogen oxides (NOx) emissions.
Optimization Considerations:
Figure E1.2a illustrates the complex trade-offs between maximizing efficiency and minimizing emissions.
Engineers must find a compromise between conflicting objectives, considering both efficiency and
environmental compliance. Additionally, the challenge of managing variable operating demands and
incorporating discrete variables further complicates the optimization process.
Note: These challenges highlight the intricate nature of optimization in boiler operations, emphasizing
the need for careful consideration and compromise in decision-making.
Example 1.3: Optimum Distillation Reflux
Historical Strategy and Energy Conservation:
Prior to 1974, distillation column trains prioritized maximizing separation (product purity) by consuming
substantial utilities like steam and cooling water, especially when fuel costs were low. However, each
column's operation has constraints, such as condenser duty, tray flooding, or reboiler duty.
Operational Challenges:
1. Problem Statement:
Challenge: Distributing a single product (Y) manufactured at multiple plant locations in large
chemical companies.
Objective: Determine production quantities at each of the m plants (Y1, Y2, ..., Ym) and
allocate the product to n demand points (Yd1, Yd2, ..., Ydn).
2. Key Considerations:
Delivery to multiple customers at varying distances from each plant.
Minimizing costs involves optimizing transportation costs between supply and demand
points and considering production cost versus capacity curves for each plant.
3. Complex Factors in Optimization:
Production Rate Variability: Plants vary in nominal production rates, and some may be more
efficient than others may.
Efficiency and Age: Newer plants may be more efficient, but older ones could be preferred if
new customers are located closer.
Unique Functionality: A unique functionality exists between production cost and rate for
each plant based on factors like efficiency and age.
Transportation Costs: Optimal solutions may involve manufacturing more from an old,
inefficient plant if it is close to new customers, despite higher costs.
Operational Constraints: Constraints on production levels from each plant affect the
distribution plan.
4. Dynamic Allocation:
Allocation decisions may need adjustment based on factors like operating rates and
transportation costs.
Excessive operation of an old plant might lead to higher costs, prompting redistribution by
other plants despite transportation costs.
5. Optimization Challenge:
Finding the cost-minimizing solution involves complex trade-offs between production costs,
transportation costs, and operational constraints.
Achieving an optimal distribution plan requires a nuanced understanding of plant
characteristics, production rates, and geographical considerations.
6. Consideration of Constraints:
Constraints on production levels from each plant add an additional layer of complexity to
the optimization problem.
7. Holistic Approach:
A holistic approach is necessary, integrating considerations of plant efficiency, production
capacity, transportation costs, and customer locations.
8. Scope of Optimization:
Beyond transportation costs, the optimization scope extends to production cost curves,
plant efficiency, and capacity constraints.
5. Essential Features of Optimization Problems
5.1 Mathematical Formulation of Optimization Problems
The solution to optimization problems involves mathematical expressions, which do not necessarily
need to be complex. This section emphasizes the use of quantitative methods and the importance of
aligning the problem statement with the chosen solution technique. The goal is to organize any process
problem for optimal solutions effectively.
Optimization problems, regardless of their complexity, share common structures. This similarity across
disciplines, such as chemical engineering, physics, and traffic engineering, allows for the development of
a universal framework. This section introduces a methodology to study any problem within a
standardized structure, involving the consideration of a process model and the selection of an objective
criterion.
A feasible solution satisfies the conditions specified by equality and inequality constraints. The feasible
region represents the set of all feasible solutions defined by these constraints. Optimal solutions are sets
of values that not only meet the conditions of categories 2 and 3 but also provide the best value for the
objective function in category 1.
If the number of unknown variables equals the sum of the independent consistent equality and
inequality constraints, at least one solution exists. However, if the number of unknown variables
exceeds this sum, the process model is termed underdetermined, leading to an infinite number of
solutions. In overdetermined models, where the equations outnumber the unknown variables, resolving
the difficulty may involve relaxing some constraints.
5.6 Notation for Optimization Problem Categories
There is no universal method or algorithm for efficient optimization across all problems. The choice of a
method depends on factors such as the nature of the objective function, the constraints involved, and
the number of variables. Six general steps are outlined for analyzing and solving optimization problems,
offering flexibility in the sequence of execution.
Table 1.1 outlines the six steps used to solve optimization problems:
i. Analyze the Process: Identify variables and specific characteristics of interest by creating a list of
all variables.
ii. Determine Optimization Criterion: Specify the objective function in terms of the variables
identified, providing the performance model.
iii. Develop Process or Equipment Model: Use mathematical expressions to create a valid model,
including equality and inequality constraints, based on physical principles, empirical relations,
and external restrictions. Identify the independent and dependent variables to establish the
number of degrees of freedom.
iv. Simplify the Problem Formulation: If the problem is too large, break it into manageable parts or
simplify the objective function and model.
v. Apply Optimization Technique: Use a suitable optimization technique to address the
mathematical statement of the problem. This often involves numerical computations using
computers.
vi. Check Answers and Examine Sensitivity: Verify the obtained solution and assess its sensitivity to
changes in coefficients and assumptions in the problem.
Steps 1, 2, and 3 focus on the mathematical definition of the problem, emphasizing the identification of
variables, the specification of the objective function, and the statement of constraints. Problem
formulation is critical, and attention to detail is given in subsequent chapters.
6.4 Simplification Strategies
Step 4 suggests simplifying the mathematical statement of the problem without losing its essence. This
can involve ignoring variables with insignificant effects, eliminating variables appearing in simple forms,
or employing strategies based on engineering judgment.
Step 5 involves applying optimization techniques to compute the optimum point. Various methods exist,
with a common reliance on computers for numerical solutions. A good initial guess is often crucial for
the effectiveness of these methods.
The final step includes checking the optimality of the solution, through either verifying sufficient
conditions or iteratively demonstrating the superiority of the obtained value. Sensitivity analysis
examines how the optimal solution responds to changes in problem parameters.
Chapter 02
Classification of Models
Models used in optimization applications fall into two broad categories, each serving specific purposes:
Many industrial processes, like the electrostatic precipitator, are too intricate for the
straightforward application of fundamental physical laws.
Thermodynamic or chemical kinetics data may be essential but might not always be
available.
Empirical models, although less effort-intensive, are often limited to specific operational
ranges and may require reformulation for different conditions.
Linear models exhibit superposition, while nonlinear models do not. Linearity is determined by
examining the equation(s) representing the process; if any term is nonlinear, the model and process are
considered nonlinear. Linear models are significantly easier to manipulate and solve than nonlinear
ones, making them more practical in many applications.
2. Steady-State versus Unsteady State
Steady-state processes, also known as time-invariant or static, maintain constant dependent variable
values over time. In contrast, unsteady-state processes, termed non-steady state, transient, or dynamic,
involve changing dependent variables with time. Most optimization problems discussed in this book are
based on steady-state models.
Continuous variables can assume any value within an interval, while discrete variables take distinct
values. In chemical engineering optimization, both discrete and continuous variables may coexist. For
instance, optimizing a compressor system involves selecting the number of compressor stages (a
discrete variable) and setting continuous variables like suction and production pressures for each stage.
Optimization problems without discrete variables are generally easier to solve than those involving
discrete variables.
Chapter 03 (Hamdy Taha)
3.1 LP Model in Equation Form
The simplex method for linear programming (LP) computations is facilitated by two requirements
imposed on the LP model:
All Constraints as Equations with Nonnegative Right-Hand Side: Ensure that all constraints are
equations with nonnegative right-hand sides. This simplifies the development of the simplex
method.
All Variables Nonnegative: All variables in the model should be nonnegative.
To convert an inequality into an equation with a nonnegative right-hand side, a nonnegative slack
variable is added or a surplus variable is subtracted. For example:
1. Slack Variable Addition: Convert the M1-constraint in the Reddy Mikks model:
6x1+4x2+s1=24, s1≥0
Here, s1 is the slack variable representing the unused amount of resource M1.
2. Surplus Variable Subtraction: Convert the 1≤2 feed mix constraint in the diet model:
Here, s1 represents the excess tons of the mix over the required minimum of 800 tons.
For unrestricted variables in an LP model, substitution with two nonnegative variables can be used. For
instance:
In this case, Si- represents the number of workers hired and Si+ the number of workers fired.
The development of the algebraic simplex method is based on the graphical LP solution. In nontrivial
LPs, where the number of equations ‘m’ is less than the number of variables ‘n’, the solution space
represented algebraically yields an infinite number of solutions, provided the equations are consistent.
The comparison between the graphical and algebraic methods is illustrated in Figure 3.1. While the
graphical solution space has an infinite number of solution points, the algebraic representation assures a
similar conclusion due to the imbalance between the number of equations and variables. For example,
the equation x+y=1 has m=1 & n=2, resulting in an infinite number of solutions represented by any point
on the straight line x+y=1.
3.3 The Simplex Method
The simplex method is an iterative optimization algorithm for solving linear programming (LP) problems.
Instead of enumerating all basic solutions (corner points) of the LP problem, the simplex method
explores a selected subset of these solutions. The iterative nature of the method involves moving from
one corner point to another until the optimal solution is reached.
1. Starting Point: The simplex method always starts at the origin, where all decision variables are
zero.
2. Objective Function Improvement: The method evaluates the objective function to determine
the variable with the largest rate of improvement. This variable is chosen to be increased first.
3. Variable Increase: The chosen variable is increased until a corner point is reached.
4. Path to Optimum: The simplex method connects corner points along the solution space in an
iterative fashion. Each corner point corresponds to an iteration.
5. Direction of Movement: The method always moves alongside the edges of the solution space,
not cutting across it.
Historical Perspective:
The simplex method was developed by American mathematician George B. Dantzig, rendering large LPs
solvable in practice. Dantzig's work was influenced by the Input-Output Economy Model by Nobel
Laureate Wassily Leontief. Initially, Dantzig faced challenges with large models and lack of a well-defined
objective function. However, he introduced the concept of an optimized objective function, leading to
the discovery of the simplex method.
Computational Details:
1. Repetitive and Tedious Computations: The simplex method involves repetitive, tedious, and
voluminous computations.
2. Computational Example: Numerical examples are used to explain the computational details of a
simplex iteration.
3. Importance of Hand Computations: While in practice, LPs are solved using computers,
experiencing hand computations is crucial for understanding the algorithm's workings. It helps
in appreciating the role of computers in solving Operations Research (OR) problems and gaining
insight into the relationship between algebraic iterations and graphical corner points.
The simplex method has stood the test of time and remains a fundamental algorithm in mathematical
optimization, playing a significant role in solving linear programming problems efficiently.
TORA Moment:
TORA (Toolkit for Operations Research and Analysis) provides an interactive and user-guided option for
solving optimization problems, particularly in the context of the simplex method. The Gauss-Jordan
computations, which are computationally intensive, tedious, and less significant in understanding the
method, are replaced with interactive user guidance in TORA.
In linear programming (LP), sensitivity analysis involves studying how changes in the parameters (input
data) of the model affect the optimal solution without altering the structure of the problem. This section
covers graphical and algebraic sensitivity analyses.
Examines how the optimal solution reacts to variations in the right-hand side of the constraints
(availability of resources).
Investigates the impact on the optimal solution when coefficients of the objective function (unit profit
or unit cost) change.
Algebraic Sensitivity Analysis – Changes in the Right-Hand Side
Extending the analysis to the general LP model, this section uses the TOYCO model as an example.
Sensitivity analysis is applied to the right-hand side of constraints, examining the dual price (unit worth
of a resource) and its feasibility ranges.
This part of the sensitivity analysis focuses on the objective function. Key concepts include:
Defines the reduced cost as the cost of consumed resources per unit revenue.
Optimality Ranges:
Determines conditions to maintain optimality, considering changes in unit revenues for various
variables.
Explores ways to transform unprofitable variables into profitable ones by increasing unit revenue or
decreasing the unit cost of consumed resources.
Heat Transfer Analysis: PA&O involves detailed analysis of heat transfer mechanisms within the heat
exchanger. This includes evaluating the effectiveness of the heat exchanger in transferring heat from
one fluid stream to another.
Pressure Drop Analysis: Examining pressure drops across the heat exchanger to optimize flow rates and
minimize energy losses.
Mathematical Modeling: Developing mathematical models to represent the heat exchanger's behavior
under different operating conditions.
Computational Fluid Dynamics (CFD): Using CFD simulations to visualize fluid flow patterns and
temperature distribution within the heat exchanger.
3. Optimization Techniques:
Thermal Efficiency Optimization: Adjusting parameters such as flow rates, temperatures, and heat
transfer surface area to maximize thermal efficiency.
Cost Optimization: Considering factors like material costs, energy costs, and maintenance costs to find
an optimal design that balances performance and expenses.
4. Energy Integration:
Pinch Analysis: Applying Pinch Technology to identify opportunities for heat recovery and optimize the
temperature profiles within the process.
Monitoring Systems: Implementing monitoring systems to detect any deviations from optimal
performance.
Control Strategies: Developing control strategies to maintain the heat exchanger's efficiency under
varying conditions.
Application in Reactors:
1. Kinetic Modeling:
Reaction Kinetics: Developing kinetic models to describe the rate of chemical reactions within the
reactor.
Reaction Mechanism: Investigating the detailed mechanisms of complex reactions to optimize reaction
pathways.
2. Process Analysis:
Mass and Energy Balances: Conducting mass and energy balances to ensure proper stoichiometry and
energy utilization.
3. Optimization Strategies:
Yield Optimization: Maximizing the yield of desired products by adjusting operating conditions.
Selectivity Optimization: Optimizing selectivity to produce the desired product with minimal undesired
by-products.
4. Catalyst Design:
Catalyst Screening: Identifying and optimizing catalysts to enhance reaction rates and selectivity.
Safety Analysis: Assessing and optimizing reactor conditions to ensure safe operation.
Environmental Impact: Evaluating the environmental impact and exploring cleaner and greener
alternatives.
Conclusion:
Process Analysis and Optimization play a crucial role in enhancing the efficiency, safety, and
environmental sustainability of heat exchangers and reactors in chemical engineering. The application of
mathematical models, simulations, and optimization techniques allows engineers to fine-tune operating
parameters and design for optimal performance. Additionally, the integration of advanced technologies,
such as artificial intelligence and machine learning, further enhances the capabilities of PA&O in
chemical engineering applications.