Guest Visit - UNIPA

8/14/2019 Guest Visit - UNIPA

http://slidepdf.com/reader/full/guest-visit-unipa 1/17

High throughput experimentsand predictive modelling

Micro-dispensing of liquid food droplets on

a solid substrate

Jimmy Perdana



Outline

High throughput experiments

Predictive modelling

Case study: thesis

Conclusions

Q&A



Project

High throughput experimentation on spraydrying

Current sub-project (thesis):Investigate dispensing of liquid micro-droplets (d = 80-150 μm) on a solidsubstrate



Why: High Throughput?

Large experiments =

• Large amount of sample• Long time experimentation

• Difficult to control• Slow respond

• More expensive



How: High Throughput?

• Small scale experiments• Can be completely different• Should be representative



Modelling

• Simplification of reality• A tool to get a grip on reality without knowing the

reality in complete detail

• Empirical model: no underlying physical theory• Mechanistic model: based on (an) underlying

physical theory

• Prediction : e.g.: shelf life• Design : e.g.: experiments• Control : e.g.: product quality



Droplet dispensing

Processparameters:

•

Valve opening time• Applied pressure• Needle tip diameter

Fluid properties:• Viscosity• Density• Rheological behavior



Droplet imaging

Imaging

( )( ) ( )( )

−−−−= α α π cos1

2

1cos1

6

1

3

4 33r V

θ α −°=180 θ ≈ 145O



Experimental result

0

50

100

150

200

250

300

350

0 0.1 0.2 0.3 0.4 0.5

ΔP (Pa)

V

( n L )

80 ms (e)

60 ms (e)

50 ms (e)

40 ms (e)

Fluid : MEGDispenser tip : 0.1 mm ID

Valveopeningtime (t)



Droplet dispensing - model

Model assumption:• Dispensing resistance is only in the

dispenser needle tip

• Steady laminar flow• No slip between fluid and needle-tip

wall• Newtonian behavior

P1

P2

x

r

L

D=2R

umax

t g L P RQt V

+∆== ρ

µ π 8

4

Poiseuille's equation



0

100

200

300

400500

600

700

800

900

1000

0 0.1 0.2 0.3 0.4 0.5

ΔP (Pa)

V

( n L )

80 ms (m)

60 ms (m)

50 ms (m)

40 ms (m)

80 ms (e)

60 ms (e)

50 ms (e)

40 ms (e)

0

100

200

300

400500

600

700

800

900

1000

0 0.1 0.2 0.3 0.4 0.5

ΔP (Pa)

V

( n L )

80 ms (m)

60 ms (m)

50 ms (m)

40 ms (m)

80 ms (e)

60 ms (e)

50 ms (e)

40 ms (e)

Model vs. Experiment

Model: underestimation at low ΔP needle-pistonmovement

overestimation at high ΔP higher resistance




Model optimization

0

4

8

V t g L

P RV

+

+

∆

=ρ

µ

π

Optimized parameters: L and V0

• L represents the total resistance of dispenser

• V0 represents the minimum volume

dispensed, caused by the piston movement



Model optimization

0

50

100

150

200

250

300

350

0 0.1 0.2 0.3 0.4 0.5

ΔP (Pa)

V

( n L )

80 ms (m)

60 ms (m)

50 ms (m)

40 ms (m)

80 ms (e)

60 ms (e)

50 ms (e)

40 ms (e)

L = 2.09 ± 0.027 cm (P = 0.95)V0 = 16.25 ± 1.08 nL (P = 0.95)


rLV0 = 0.71



Model optimization

-60

-40

-20

0

20

40

60

80

0 0.1 0.2 0.3 0.4 0.5

ΔP (Pa)

V r e s i d u a l [

V m o d e

l - V e x p . ]

( n L )

80 ms

60 ms

50 ms

40 ms

Residual is scattered, confidence interval P =0.95




Effect of fluid viscosity

Model “confirms” the effect of fluidviscosity

0

20

40

60

80

100

120

140

160

180

0 0.1 0.2 0.3 0.4 0.5

ΔP (Pa)

V

( n L )

DEG

PEG



Conclusion

•

High throughput experiments andpredictive modelling can be used as thetools to improve the research



Thank you

Questions?

Guest Visit - UNIPA

Documents