You are on page 1of 12

Thermodynamic and Flow Calculations for the Design of a

Cold Air Blower

Brandon Boiko, Mike Snow


Summary: The purpose of this document is to provide a detailed explanation for the design
changes to the blower system proposed by HPD. Most calculations are based on or adapted from
calculations performed by Daniel Hofstadter in his DKIST GOS Enclosure Environmental
Control System Analysis.

Design Requirements:
1. Air temperature rise of GOS box not to exceed 9𝐾.
2. Surface temperature of components maintained between -3C to +1.5C from ambient during
operating conditions.
3. Positive pressure within the GOS box to blowout dust while maintaining primary flow
through the return lines.

Assumptions:
1. Steady state, steady flow conditions
2. Well insulated/adiabatic conditions everywhere except at heat exchangers
3. Incompressible, constant air properties
4. Uniform pressure inside GOS box

Air Properties:
𝑃𝑎𝑡𝑚 = 69.8 𝑘𝑃𝑎
𝑇𝑎𝑚𝑏 = 295 𝐾
𝜌𝑎𝑖𝑟 = 0.82 𝑘𝑔/𝑚3
𝑘𝑎𝑖𝑟 = 0.02624 𝑊/𝑚 ∙ 𝐾
𝐶𝑝𝑎𝑖𝑟 = 1004 𝐽/𝑘𝑔 ∙ 𝐾
𝜇𝑎𝑖𝑟 = 1.81 × 10−5 𝑘𝑔/𝑚 ∙ 𝑠

Analysis: This analysis is broken into a few parts. First, the minimum flow rate to maintain the
temperature inside the GOS box below 9𝐾 above ambient is determined. This requirement, as
described in Hofstadter’s analysis, is set for safety reasons, but the exact origin of the 9𝐾 value is
still unclear. It is speculated that it may be derived by a limit set by the chiller, where the
minimum set point for the process fluid is ambient−9𝐾. To keep calculations consistent with
Hofstadter’s analysis, we will use this as the maximum temperature change allowed for the air.
Selected blower components will be compared with the calculations to verify they can provide
the system requirements. The second part describes the reasoning behind removing the second
fan and provides calculations to support this decision. Next, we will look at the maximum
diameter dust particle that the expelled air will prevent from entering the GOS enclosure. Finally,
we will look at what the maximum temperature will be during calibration, and determine what
effects additional fans might have.
An updated diagram of the new blower design is shown in Figure 1. This differs from the
previous design by the removal of the return fan prior to the intake for replacement air.
Additionally the “Optional” second heat exchanger is not shown. The second heat exchanger was
suggested as an option to accommodate larger headloss in the GOS box since the exact flow
characteristics inside the GOS box are hard to determine. Larger headloss would cause the
flowrate to decrease, meaning a reduction in the cooling power of the heat exchanger.

Figure 1: Layout of thermal rack blower. For reference 206𝑊 max is the heat load of the M5
system which is larger than the GOS box under operational conditions.

Power dissipation
A minimum air flow required to maintain temperatures inside the GOS box below 9𝐾 above
ambient can be calculated by solving the first law equation for the volumetric air flow.

𝑄̇ = 𝜌𝑉̇ 𝐶𝑝∆𝑇
Rearranged:
𝑄̇
𝑉̇ =
𝜌𝐶𝑝∆𝑇

Here the heat transfer is just the heat load of the GOS box, 50𝑊. With a safety factor of 2 and a
minimum increase in temperature of 9𝐾, the minimum flow rate required to maintain safe
temperatures under operating conditions is:

𝑉̇ = 0.0135 𝑚3 /𝑠
This value is approximately 𝟓𝟎𝒎𝟑 /𝒉𝒓, or about 𝟐𝟕 𝑪𝑭𝑴. Figure 2 shows the performance
curve of the ebmpapst NRG118/0800-3612 blower. We will revisit this diagram later once we
calculate the headloss throughout the system after accounting for blowout or air leakage, but for
now note that the blower can provide 𝟓𝟎𝒎𝟑 /𝒉𝒓 at a static pressure of 𝟏𝟗𝟎𝟎 𝑷𝒂.

Figure 2: Performance curve for the ebmpapst NRG118/0800-3612 blower. This blower operates
at 8500 𝑟𝑝𝑚 with a power input of 61𝑊.

Figure 3 shows the performance curve for the Lytron 6310 heat exchanger. The flow rate for our
system is rather low which puts us on the left side of the performance curves. While in this
region the curves may not be as accurate, Hofstadter assumed a value of 20 𝑊/𝐾 for the heat
exchanger if water was circulated at 3 𝑙𝑝𝑚.

𝑪𝑶𝑷𝑯𝑬𝑿 = 𝟐𝟎 𝑾/𝑲

Note the temperature in the above value is in units of degrees temperature difference between the
inlet air and the inlet water. This means that units of K can be used instead of C to keep things
consistent with the rest of the analysis. Additionally, it is very important to use the initial
temperature difference to calculate the cooling power.

Figure 3: Performance curve for the Lytron 6310 heat exchanger.

Although it is believed that one fan is adequate to provide the flow for the system, the cooling
system will be analyzed using a two fan heat load to keep calculations consistent with
Hofstadter’s analysis. If 172 𝑊 is absorbed by air, then the increase in temperature can be
determined by the heat equation.

∆𝑻 = 𝑸̇/𝝆𝑽̇𝑪𝒑

Now the heat transfer for the entire system is the heat load of the GOS box plus the heat load of
the fans, for a total of 172𝑊. Plugging in values yields the overall temperature gradient of.

∆𝑻 = 𝟏𝟓. 𝟒𝟖 𝑲
This does not account for heat lost due to blowout, or heat gained through fresh intake air. The
actual temperature of the return air will likely be lower due to these reasons. We can then
multiply this value by the coefficient of performance for the heat exchanger to get the cooling
power of the heat exchanger.
𝐶𝑂𝑃𝐻𝐸𝑋 ∆𝑇 = 310 𝑊

This value is larger than the heat load of 172 𝑊 which means the heat exchanger is capable of
cooling the full 172 𝑊. Note that an increase in the return air temperature will also increase the
cooling power of the heat exchanger. This is because a larger temperature gradient increases the
overall heat transfer. As long as the process fluid temperature can be maintained, the heat
exchanger is capable of removing the heat. This means that the constraining component of the
system is the chiller. If the chiller cannot handle the heat load the process fluid temperature will
increase until a thermal equilibrium is reached. If water absorbs the full 172𝑊 the temperature
of the water will increase by 2.42 𝐾. If we assume the outlet air temperature will leave the heat
exchanger at the average temperature of the inlet and outlet water and the outlet air must leave at
ambient temperature, then the inlet water stream should be maintained at ambient −1.21 𝐾.
Figure 4 shows the performance curve for the thermocube 600L.

Figure 4: Performance curve for thermocube 600L, shown in orange.


If the process fluid is to be held at ambient −1.21𝐾 and the facility water is held steady at
ambient −4𝐾, that is a temperature difference of 2.8𝐾. Since the process fluid is being held
at 2.8𝐾 above the facility water, the thermocube 600L can provide approximately 750𝑊 of
cooling power.

Flow/Pressure Calculations
The pressure calculations can be simplified by utilizing a resistor analogy. Figure 5 shows a
diagram of the flow resistive circuit. R1 is the supply resistance, R2 is the return resistance, R3 is
the orifice resistance of the upper and lower windows when covers are in the open position, and
R4 is the resistance of the air intake for replacement air.

Figure 5: equivalent resistance circuit for the air flow system.

From Hofstadter’s analysis the resistances of R1 and R2 can be calculated by using Poiseuille’s
law along with the headloss values for the supply and return lines.

∆𝑃
𝑅=
𝑉̇

The pressure inside the GOS box is determined by how much air we want to allow to leak out of
the orifices. For the purpose of this analysis we can assume no more than 10% of the air flow
rate is allowed to leak out of the GOS box. Using the orifice headloss equation we can calculate
the pressure inside the GOS box. The upper and lower orifice diameters are 170mm and 112mm
respectively.

𝜌𝑉̇𝑙𝑒𝑎𝑘 = .60𝐴𝑡 √2𝜌𝑃2

Solving this for P2 yields a small pressure of:

𝑃2 = 0.0017 𝑃𝑎
Now that we know the pressure inside the GOS box we can calculate the required pressures at
points 1 and 3.
𝑃1 = 𝑉̇ 𝑅1 + 𝑃2
𝑃3 = 𝑃2 − 𝑉̇𝑟𝑒𝑡𝑢𝑟𝑛 𝑅2

Here the return flow rate is calculated similarly to the leak flow rate, assuming steady state
conditions and continuity holds. The return flow must be 90% of the air flow rate. We can now
find the difference in the pressures at point 1 and point 3. This is the static pressure difference for
the fan and it has a value of about 663 𝑃𝑎. Looking back at Figure 2, the fan is more than
capable of providing a flow rate of 50 𝑚3 /ℎ𝑟 at that static pressure. Since the fan is not being
controlled, the actual flow rate will likely be higher than the minimum, and will reach
equilibrium once the headloss increases to a point where flow rate and static pressure meet at a
point on the performance curve of the fan.

Lastly, the leak flow rate and the GOS box pressure can be controlled to some extent by
changing the resistance of the air intake, R4. A damper or butterfly valve could be used to
constrict, or relieve the flow and achieve different leak flow rates. Note there are theoretical
limits to how much the flow can be controlled. If the leak flow was totally constricted, R4 goes
to infinity, then the GOS box pressure drops to ambient, and dust particles will be allowed to
blow into the box. If the intake is fully open, R4 goes to zero, then 100% of the flow is blown out
of the GOS box and the pressure becomes 0.17 𝑃𝑎. This is the theoretical maximum pressure
drop across the orifice with a flow rate equal to the flow rate supplied by the fan.

Dust Particle Analysis


Here, we will determine whether the jet velocity exiting the windows is enough to blowout
average sized dust particles. This problem can be modeled as a spherical dust particle falling at
terminal velocity, which is then met with a jet of air. Our goal is to determine the maximum
diameter of dust particle that the jet of air can stop.

First, we must calculate the velocity of the jet of air exiting the window orifice. We can assume
that the jet is mostly uniform due to its large diameter. Then, we can equate the orifice headloss
equation to the continuity equation and solve for the velocity. This yields the following equation:

𝐾√2𝜌𝑃2
𝑢=
𝜌

From this equation we can see that the area of the orifice does not factor into the velocity at all.
Only the pressure differential affects the jet velocity, therefore a hole of any size in the enclosure
boundary will produce the same jet velocity. For a 10% leak rate, this produces a jet with a
velocity of 3.86 𝑐𝑚/𝑠.

The next step is to determine the maximum diameter of a particle that this jet can stop. We can
assume that the heaviest dust particle will be primarily made of silicon, which has a density of
about 2500 𝑘𝑔/𝑚3. Due to the scale of the particles, we can assume the kinematics to occur
quickly. Under these conditions, an air jet at the terminal velocity of the dust particle will be
enough to stop the particle from entering the enclosure. For a small particle and at low velocities,
viscous drag dominates over form drag. To calculate the terminal velocity of the particle we can
use Stoke’s law of viscous drag. The equation for terminal velocity comes from equating the
force of gravity with the viscous drag force and the buoyant force.

2 𝜌𝑑𝑢𝑠𝑡 − 𝜌𝑎𝑖𝑟 𝐷 2
𝑢= 𝑔( )
9 𝜇𝑎𝑖𝑟 2

We can rearrange this to solve for the diameter, D.

𝐷 = 2√9𝑢𝑎𝑖𝑟 𝜇𝑎𝑖𝑟 /(2(𝜌𝑑𝑢𝑠𝑡 − 𝜌𝑎𝑖𝑟 )𝑔

A dust particle moving at terminal velocity equal to the jet velocity will have a diameter of
approximately 22.7𝜇𝑚. This number can be increased by adjusting the flow characteristics such
that more air is allowed to blow out of the GOS box.

Calibration Temperatures
Under calibration conditions the heat load at the GOS box will increase greatly. The calibration
lamp has a power load of 600𝑊. Some of the energy may leave the box as reflected light, but
let’s assume all of the power goes into heating the cooling air. Using the heat equation with the
new GOS heat load of 650𝑊, we can calculate the change in temperature after the box. Under
these assumptions the air will raise in temperature by 58.5𝐾. If air enters at ambient, then it will
leave at 353.5𝐾, or about 80𝐶.

𝑇𝑎𝑖𝑟,𝑐𝑎𝑙 = 80𝐶

This value is the temperature that the air leaves after it accepts the entire 650𝑊 of the GOS box.
The temperatures of components local to the lamp will likely reach higher values, until the heat
transfer leaving those components equals the heat load supplied by the lamp. Let’s look at the
steady state temperature of the lamp baffle during calibration. Since the highest temperature of
the air is 80𝐶 and the air enters at 22𝐶, there exists a temperature gradient in the air as it travels
downstream. To simplify this analysis we can perform our heat transfer calculations at the
average air temperature of 51𝐶. A majority of the heat will be absorbed by the lamp baffle.
Assuming a worst case scenario where all 650𝑊 goes to heating the baffle, we can then
calculate the temperature of the baffle at thermal equilibrium. The primary modes of heat
transfer for this system are convection and radiation. There is conduction, but the baffle is well
insulated and we can obtain a more conservative value if we neglect conduction. With these
assumptions the energy balance equation looks like:

𝑄̇ = ℎ𝐴(𝑇𝑠𝑢𝑟𝑓 − 𝑇∞ ) + 𝜎𝜀𝐴(𝑇𝑠𝑢𝑟𝑓
4
− 𝑇∞4 )

Solving for the surface temperature numerically yields a value of 228𝐶. Keep in mind this is a
worst case scenario. The actual temperature would be lower if we included heat transferred
through conduction. The heat transfer coefficient used was 5 𝑊/𝑚2 ∙ 𝐾, this value is based on
similar values for natural convection in air. Due to the size of the GOS box and the low flow
rates, it is expected that the lamp and any other component within the box would experience very
little flow velocity. If we take the cross sectional area of the internal GOS box, and apply a 20%
fill factor, then the flow area within the box is 0.3882 𝑚2. Applying continuity gives us an
average flow velocity of 1.7 𝑐𝑚/𝑠 within the box. Heat transfer coefficients for low flow forced
convection are only accurate to speeds as slow as 2 𝑚/𝑠 .

One thing to consider is since the fan is not being controlled, it will operate at maximum rpm.
Therefore, the flow rate the fan provides will be equal to the flowrate at which the headloss
through the system equals the static pressure capabilities of the fan. Figure 6 shows the
intersection between the resistance curve, which is assumed to be linear i.e. constant resistance,
and a fitted performance curve for the fan. The two lines intersect at a flow rate of 76.16 𝑚3 /ℎ𝑟
with a static pressure of 1048 𝑃𝑎. This new flow rate can be applied to the previous calculations
to obtain the actual performance of the system. These values are located in Table 1.

Table 1: Calculated properties at the actual flow rate


Property Value Units
𝑽̇ 76.16 𝑚3 /ℎ𝑟
𝑪𝑶𝑷𝑯𝑬𝑿 30 𝑊/𝐾
∆𝑻𝒐𝒑𝒆𝒓𝒂𝒕𝒊𝒐𝒏𝒂𝒍 9.88 𝐾
𝑸𝑯𝑬𝑿 296.4 𝑊
𝑷𝟐 0.0042 𝑃𝑎
𝑽𝒍𝒆𝒂𝑘,10% 7.62 𝑚3 /ℎ𝑟
𝒖𝒋𝒆𝒕 6.05 𝑐𝑚/𝑠
𝑫𝒎𝒂𝒙 28.37 𝜇𝑚
𝑻𝑎𝑖𝑟,𝒄𝒂𝒍 59.32 𝐶
𝑻𝒍𝒂𝒎𝒑 224.18 𝐶
4000
System Resistance
3500 Single Fan
Series Fans
3000 Parallel Fans

2500
Static Pressure (Pa)

2000 X: 121.3
Y: 1640

X: 85.55
1500 X: 76.16
Y: 1153
Y: 1048

1000

500

-500
40 60 80 100 120 140 160 180 200
3
Flow Rate (m /hr)

Figure 6: Fan performance curves (orange) intersects the resistance curve (blue) at the actual
flow conditions of the system

After implementing the larger flow velocity we do notice some improvement in the performance
of the system. However, the small increase in performance compared with the relatively large
increase in flow (+50% of the original) does not look promising. We can look at the effect of
adding fans in series, or in parallel by modifying the performance curve. Fans in series doubles
the static pressure capabilities without changing flow rate, or shifts the curve vertically. Fans in
parallel doubles the flow rate without changing static pressure capabilities, or shifts the curve
horizontally. Figure 6 also shows the intersection points for both fans in series and fans in
parallel. The parallel case increases the overall flow rate more than the series case. Table 2
shows the performance of the parallel case. It is clear from these calculations that increasing the
flow rate by adding more fans offers very little change in the performance. A single fan is
capable of handling operational conditions, and increasing the flow rate does not reduce the
maximum temperatures significantly during calibration conditions.
Table 2: Calculated properties at the flow rate provided by two fans in parallel.
Property Value Units
𝑽̇ 116.7 𝑚3 /ℎ𝑟
𝑪𝑶𝑷𝑯𝑬𝑿 35 𝑊/𝐾
∆𝑻𝒐𝒑𝒆𝒓𝒂𝒕𝒊𝒐𝒏𝒂𝒍 6.44 𝐾
𝑸𝑯𝑬𝑿 225.4 𝑊
𝑷𝟐 0.0098 𝑃𝑎
𝑽𝒍𝒆𝒂𝑘,10% 11.67 𝑚3 /ℎ𝑟
𝒖𝒋𝒆𝒕 9.27 𝑐𝑚/𝑠
𝑫𝒎𝒂𝒙 35.12 𝜇𝑚
𝑻𝑎𝑖𝑟,𝒄𝒂𝒍 46.36 𝐶
𝑻𝒍𝒂𝒎𝒑 221.68 𝐶

You might also like