Mathematical Modeling

Mathematical Models
3rd classification
on the base of
• mathematical features of equations and variables,
• also affecting type or easiness of solution
 Ch.3 of Himmelblau D.M. and Bischoff K.B., “Process Analysis and
Simulation”, Wiley, 1967
1. deterministic vs stocastic
2. lumped parameters vs distributed parameters
3. one dimension vs several dimensions
4. linear vs non linear
5. steady state vs time dependent
6. time-invariant vs time – varying
7. autonomous vs non – autonomous

Dynamical models
2
• A deterministic model is a model where:
– the material properties are well known, i.e. deterministic.
none of them is random
– The applied load are also deterministic
– Examples: timetables, pricing structures, a linear
programming model, accounting
• A Stochastic/Probabilistic model has on the other hand:
– random properties, e.g. the Young's modulus is a random
variable with uniform distribution [E1, E2]; or normal
distribution (of a given mean or standard deviation)
– The applied load is random variable, e.g. Wind Load,
earthquake (vibration of random amplitude and
displacement)
• The Hybrid model is a "mixture" of both Deterministic and
Stochastic. Its treatment is quite similar to the Stochastic
model. The presence of a single random variable in the model
necessitates the consideration of the stochastic treatment. 3
 Distinguishing Features of Lumped and Distributed
Models

• Physics
– Does point rainfall-runoff model represent well ‘field’ processes
– Can hillslope/channel routing be represented well on ‘practically
reasonable’ space/time scales
– Does statistical approach solve a basin heterogeneity problem

Lumped Distributed
 Distinguishing Features of Lumped and Distributed
Models (Continued)

• Physics
– Does statistical approach solve a basin heterogeneity problem
0

-20

-40

-60

-80

-100
1E1 1E2 1E3 1E4 1E5
Grid scale, sq. km.

SAC-SMA Reformulated SAC-SMA

SWB

Surface runoff simulated with and without use of

rainfall distribution function at different scales
• Space/Time Variability
– Does accounting for the space/time variability of input data and
parameters guarantee better results
– Does scale effect significantly on the model structure
– Is a lumped model a reasonable candidate in a distributed system

30
Effect of noisy rainfall data on the peak volume
at different simulation scales.
25

20

15

10

0
1 10 100
Number of pixels per basin

Noise=0% Noise=25% Noise=50% Noise=75%

• Parameterization/Calibration
– Can distributed model parameters be measured on the grid scale
– Are distributed model parameters identifiable enough from
hydrograph analyses
– How much does scale effect on model parameters

3.5
Change an ‘effective’ parameter value at different
scales as a function of rainfall variability
3

2.5

1.5

1
0 10 20 30
Relative scale, L (k)/L(0)
 Mathematical Models
3rd classification
on the base of
• mathematical features of equations and variables,
• also affecting type or easiness of solution
 Ch.3 of Himmelblau D.M. and Bischoff K.B., “Process Analysis and
Simulation”, Wiley, 1967

1. deterministic vs stocastic
2. lumped parameters vs distributed parameters
3. one dimension vs several dimensions
4. linear vs non linear
5. steady state vs time dependent
6. time-invariant vs time – varying
7. autonomous vs non – autonomous

9
 Mathematical Models
4th classification
(based on the type of description adopted for time
as the independent variable)

 VALID for DYNAMICAL SYSTEMS only

1. continuous time models

2. discrete time models

10
 Mathematical Models
5th classification
(based on the type of solution,
with exclusion of empirical models)

1. ANALYTICAL SOLUTIONS
 DIFFERENTIAL CALCULUS
 ALGEBRA
2. NUMERICAL SOLUTIONS
 NUMERICAL CALCULUS
Ex.: numerical methods for PDEs

1. FINITE DIFFERENCE METHODS (FDM)

2. FINITE ELEMENT METHODS (FEM)

11
3. COLLOCATION METHODS