|
 |
[ Pinch Analysis ] [ Process simulation ] [ Electricity savings ] [ Product optimization ] [ Rental ] [ Training ]
CUSTOM MATHEMATICAL
MODELIZATION
AND OPTIMIZATION OF
PRODUCTS, EQUIPMENTS, UNIT OPERATIONS, ETC. |
| |
|
  Pragmathic offers a very
pertinent mix of expertise that can be used to customize solutions to your
situation. We can help you optimizing many aspects of your industrial activities
and increasing your plant's profitability and competitiveness.Whether
you need to optimize a product (weight, volume, performance, cost,
temperature rise, ...), improve a unit operation or equipment,
troubleshoot your process or identify variability controlling factors,
our knowledge in fundamental engineering sciences, mathematics and
statistics can deliver
cost-effective high fidelity models and solutions. In some cases, our
work has led to patents.
If you need to simulate and
optimize something unusual, we can help you!
Fundamental
engineering sciences
Transport phenomena (momentum, heat,
mass, diffusion)
Thermodynamics
Chemical and biochemical reaction kinetics
Physic
Rheology
Simulation
numerical
methods
Analytical developments
Finite elements, volumes and differences (ex.: CFD)
Statistical multivariate analysis
Dimensional analysis
Surface response methodology
Data
acquisition methods
Lab + design of experiments
Process + DOE
Operation databases
Optimization
methods
Marquardt
Generalized reduced gradient
LP, QP, MILP, MINLP
Dynamic programming
|
Examples
of tailored applications(1)
Thermodynamic
model of the solid Lithium-Polymer battery technology (download here) (2)
Finite
volume heat, mass and momentum transfer in a steam
reboiler (capacity maximization)
(3)
Computer simulation aided design of a calorimeter able to measure heat rates of 0.5 to 10 W within 0.1W.
(4)
Statistical prediction model (PLS) of a TMP pulp freeness
|
(5)
Stress in a pressed Lithium-Polymer cell (MSC Nastran for Windows)
| |
|
 Application
example: design optimization of a new automatic door operator
Step 1: Using design of experiments, find
design weaknesses vs life span and noise. Factors respectively found:
bearing and chain sizes.
 Step 2:
Develop a life span model to meet design specs (100,000 cycles),
minimize components costs and support manufacturing process quality
control. Ex.: model predicted a reject rate of X% at 95% confidence
level with no change on the original components selection.
|
|
Design
of experiments (DOE) |
We design custom experimental programs
using two different approaches as required:
 |
the extensive Montgomery's
Response Surface Methodology, in which the Taguchi methods are a
special case,
|
|
the Dorian Shainin techniques
especially developped to solve quality problems in manufacturing
processes. |
|
| |
|
| |
|
|
Multivariate
Multivariate
statistical analysis |
Many multivariate analysis methods are
available: partial least squares (PLS), principal component analysis and
regression, general linear or non-linear (Marquardt) multiple regression
analysis, and dimensional analysis (an often neglected approach).
We have developped special tools to
handle cases where non-linear effects and variable time lags happen. We
have also developped means to check the level of information losses in
compressed archived data. Go here for more
information. |
|
| |
|
Main
commercial
softwares used:
Umetri Simca-p and Modde (multivariate statistical analysis and DOE), Visual basic,
Matlab, MSC Nastran for Windows (finite element analysis).
|
| |
|
|
