46
3 Simulating Droplet Microfluidic Networks
respect to their robustness. The results show a clear preference for the specification
with ID 2. Exactly this specification is the one which was eventually realized
in [13] (cf. part “Prototyping and Testing” on page 41). The appealing features of
this simulator are clear: Instead of six fabricated prototypes, one person month of
manual labor, and a total of USD 1200 of further costs, utilizing the simulator would
allow for obtaining the same result by only fabricating one single design, spending
only 1/6 of the previously spent time of manual labor plus few hours for simulating,
and a total of USD 200. Apparently, this is a significant improvement compared to
the design process applied in [13].
Utilizing the Simulator for Design Exploration
As discussed on page 42 (in part “Further Missed Potential”), the complexity of the
design process is not only a burden to get a design realizing the desired behavior,
but also prevents the designer from exploring even better solutions. Utilizing
the simulator, this burden is significantly reduced. In fact, as described next, all
questions raised on page 42 (in part “Further Missed Potential”) can now efficiently
be addressed.
Question 1: What is the minimal bypass channel length?
Here, designs are explored with respect to the limits of the bypass channel length.
Recall, a short bypass channel length decreases the time for a droplet to be trapped
and, hence, affects the throughput of the design. Furthermore, a short bypass channel
length also minimizes the area of the physical design, which is an important criterion
of the design due to the limited space of a typical microfluidic device.
The bypass channel lengths determine the volumetric flow rate into the bypass
Q bypass and also into the trapping well Q trap . Therefore, the bypass channel length
is limited by Objective 1 (cf. Sect. 3.4.1), i.e. the volumetric flow rate into an empty
trapping well has to be larger than the volumetric flow rate into the bypass channel.
In order to explore the limits, simulations are conducted where the bypass
channel length is stepwise reduced until the first droplet does not enter an empty
trapping well (which violates the objective Q trap > Q bypass ). For a gap width of
w gap = 15 μm and for a gap width of w gap = 25 μm, the simulation predicts a
minimal bypass channel length of L bypass = 2800 μm and L bypass = 900 μm,
respectively. Both values are only theoretical limits and respective designs would be
sensitive to any imperfections as, e.g., dust or imprecisions caused by the fabrication
process. Therefore, a prototype will never be pushed to these limits. Instead, these
limits allow the designer to estimate the robustness of the design. For example, this
design exploration would have prevented a prototype with L bypass = 3000 μm and
w gap = 15 μm (i.e., the specification with ID 1 from Table 3.3) to be considered in
the first place.
Question 2: What is the maximum pressure over five sets of trapping wells so that
no objective is violated?
3 Simulating Droplet Microfluidic Networks
respect to their robustness. The results show a clear preference for the specification
with ID 2. Exactly this specification is the one which was eventually realized
in [13] (cf. part “Prototyping and Testing” on page 41). The appealing features of
this simulator are clear: Instead of six fabricated prototypes, one person month of
manual labor, and a total of USD 1200 of further costs, utilizing the simulator would
allow for obtaining the same result by only fabricating one single design, spending
only 1/6 of the previously spent time of manual labor plus few hours for simulating,
and a total of USD 200. Apparently, this is a significant improvement compared to
the design process applied in [13].
Utilizing the Simulator for Design Exploration
As discussed on page 42 (in part “Further Missed Potential”), the complexity of the
design process is not only a burden to get a design realizing the desired behavior,
but also prevents the designer from exploring even better solutions. Utilizing
the simulator, this burden is significantly reduced. In fact, as described next, all
questions raised on page 42 (in part “Further Missed Potential”) can now efficiently
be addressed.
Question 1: What is the minimal bypass channel length?
Here, designs are explored with respect to the limits of the bypass channel length.
Recall, a short bypass channel length decreases the time for a droplet to be trapped
and, hence, affects the throughput of the design. Furthermore, a short bypass channel
length also minimizes the area of the physical design, which is an important criterion
of the design due to the limited space of a typical microfluidic device.
The bypass channel lengths determine the volumetric flow rate into the bypass
Q bypass and also into the trapping well Q trap . Therefore, the bypass channel length
is limited by Objective 1 (cf. Sect. 3.4.1), i.e. the volumetric flow rate into an empty
trapping well has to be larger than the volumetric flow rate into the bypass channel.
In order to explore the limits, simulations are conducted where the bypass
channel length is stepwise reduced until the first droplet does not enter an empty
trapping well (which violates the objective Q trap > Q bypass ). For a gap width of
w gap = 15 μm and for a gap width of w gap = 25 μm, the simulation predicts a
minimal bypass channel length of L bypass = 2800 μm and L bypass = 900 μm,
respectively. Both values are only theoretical limits and respective designs would be
sensitive to any imperfections as, e.g., dust or imprecisions caused by the fabrication
process. Therefore, a prototype will never be pushed to these limits. Instead, these
limits allow the designer to estimate the robustness of the design. For example, this
design exploration would have prevented a prototype with L bypass = 3000 μm and
w gap = 15 μm (i.e., the specification with ID 1 from Table 3.3) to be considered in
the first place.
Question 2: What is the maximum pressure over five sets of trapping wells so that
no objective is violated?