Professional Documents
Culture Documents
Riezqa Andika
Why Modeling?
Problem definition
▪ Defining the problems involves stating
clear goals for the modeling, including
the various elements that pertain to the
problem and its solution
▪ Questions:
• What is the objective (i.e. what
questions should the model be able to
answer)?
• What resolution is needed?
• What degree of accuracy is required?
Model Development
Conceptual model
▪ Stating hypothesis and assumptions that
are applied for the mathematical model
Mathematical model
▪ What are the variables (dependent,
independent, parameters)?
Solution
▪ Analytical
• Limited applications
• Approximate solution
• Wide applications
Estimation of parameters
▪ Appropriate values must be used in
the model
▪ Sensitivity analysis
Model Development
Application
The validated model is then ready to be used
for one or several purposes as described
earlier, e.g. to enhance our understanding,
make predictions, and give information about
how to control the process.
How Good is the Model?
Linear
▪ linear models are linear in the parameters which have
to be estimated, but not necessarily in the independent
variables (typically in statistics, regression).
▪ The computation simpler for linear problems than for
nonlinear ones.
Nonlinear
Batch vs Continuous
Batch
▪ Feedstocks for each processing step (i.e., reaction, distillation)
are charged into the equipment at the start of processing;
products are removed at the end of processing
▪ Transfer of material from one item of equipment to the next
occurs discontinuously – often via intermediate storage tanks
▪ Batch processes are intrinsically dynamic – conditions within
the equipment vary over the duration of the batch
Continuous
▪ Involve continuous flows of material from one processing unit
to the next
▪ Usually designed to operate at steady-state; due to external
disturbances, even continuous processes operate dynamically
Batch vs Continuous
Batch (Blending)
Continuous (Blending)
Steady State vs Non-steady State
Lumped-parameter vs Distributed-parameter
FA 0 − FA 𝐹𝐴
𝑑𝐹𝐴
V= 𝑉=න
−rA 𝐹𝐴0 𝑟𝐴
Lumped-parameter vs Distributed-parameter
Lumped-parameter vs Distributed-parameter
Lumped
(Almost) perfect mixing – at any particular time instant, the
values of operating conditions are (approximately) the same
at all points within the unit
Distributed
Imperfect mixing will result in different operating conditions
at different points even at the same time → existence of
distributions of conditions over spatial domains
Lumped-parameter vs Distributed-parameter
Lumped
▪ Characterized by a single independent variable (time)
▪ Their modelling can be effected in terms of ordinary
differential equations (ODEs)
Distributed
▪ Introduce additional independent variables (e.g., one or
more spatial co-ordinates, particle size, molecular weight,
etc.)
▪ Involves partial differential equations (PDEs) in time
Continuous vs Discrete
Deterministic vs Stochastic
Deterministic model
Stochastic model
Interpolation vs Extrapolation
Mechanistic vs Empirical
Mechanistic
▪ Developed from physical laws
▪ Lower dependence on range of data
▪ Complex to solve
Empirical
▪ Developed by correlating dimensionless numbers
▪ Dependence on range of (experiment) data
▪ Extrapolation is uncertain and interpolation can be very
good
▪ Simple to solve
Mechanistic vs Empirical
Ease of Solution Using Analytical Methods
Mathematical Complexity
Degree of Resolution (Scale)
Model Formulation Based On Physicochemical Phenomena
▪ Mass Balance
▪ Energy Balance
▪ Transport Phenomena
General Balance:
Energy Balance:
Balance and Conservation Principles
▪ Mass Balance
▪ Energy Balance
▪ Transport Phenomena
General Balance:
Energy Balance:
Transport Phenomena Models
driving force
transport rate =
transport resistance
transport rate
flux = normalised transport rate =
transport area
▪ Boundary conditions
▪ Population balance models
▪ Residence time distribution (RTD) theory