How Effective Are Mixing Boxes At Performing the Economizer Function?

ASHRAE JOURNAL ashrae.org DECEMBER 202022

How Effective

Are Mixing Boxes

At Performing the

Economizer Function?

BY STEPHEN WELTY, P.E., MEMBER ASHRAE; RICARDO ORTIZ

Taking outdoor air into the building to help cool a space and provide required venti-

lation is a well-known technique; to do it to maximum effect takes careful analysis

and proper selection of components and control strategies. A mixing box carries out

the economizer function of providing cooling assist and ventilation in buildings by

taking in outdoor air in a single, packaged device. This article examines the effective-

ness of using a mixing box for air mixing and explores strategies for improving its

effectiveness.

Economizers and mixing boxes both consist of a

damper to let outdoor air into the system, a damper to

let return air mix with that outdoor air and an exhaust

damper to prevent over-pressurization of the building.

Mixing boxes can be implemented with all the damp-

ers in one product (built-in relief) or with the relief or

exhaust damper separate. An economizer diagram is

shown in Figure 1.

The main advantage of a mixing box is that it can

be installed as a single unit right before the air han-

dler return and can be implemented with ﬁlter racks

and ultraviolet lights to provide air ﬁltration and

sterilization. Since the dampers are close to each other,

they can be mechanically linked and use a single actua-

tor, which reduces cost, simpliﬁes the control strategy

and reduces potential failure points. Since all the econo-

mizer components are in a single location, maintenance

and periodic performance checks are easier.

Dampers installed in ductwork in an economizer

setup can allow for long sections of duct for full mixing

to occur before the cooling coil. Mixing boxes, how-

ever, usually have short sections before the cooling coil,

which can be a disadvantage for effective air mixing. In

worst-case conditions, poor mixing of airstreams could

Stephen Welty, P.E., is vice president, and Ricardo Ortiz is design engineer at Plenums of Florida near Tampa Bay, Fla.

TECHNICAL FEATURETECHNICAL FEATURE

be copied and/or distributed electronically or in paper form without permission of ASHRAE. For more information about ASHRAE Journal,

visit www.ashrae.org.

DECEMBER 2020 ashrae.org ASHRAE JOURNAL 23

lead to coil freezing.

The other effect of poor mixing is

causing the economizer controls to operate to the detri-

ment of system efﬁciency since the sensors are sens-

ing nonrepresentative conditions.

ASHRAE Standard

90.1-2019

makes recommendations for the high limit

settings in various climate regions, but this assumes a

representative mixed air temperature is sensed.

Several CFD simulations were run to calibrate the

CFD model to test data published in ASHRAE Research

Report RP-1045, “Verifying Mixed Air Damper

Temperature Control and Air Mixing Characteristics.”

by Finaish, et al.

The report tested a wide range of mix-

ing box conﬁgurations and damper sizes, so a single

test conﬁguration was chosen that was close in size and

conﬁguration to a commercially available mixing box

for use with a 20 ton (70 kW) to 25 ton (87.5 kW) air han-

dler. The test setup is shown in Figure 2. Figure 3 shows the

commercial mixing box product. The test data available

is for a ﬂow rate of 16,000 cfm (7551 L/s) at 15% outdoor

air (OA) and 30% OA. This paper focuses on the 30% OA

case.

The steps used to proceed from the validated CFD

simulation against test data of the setup in Figure 2 to the

commercial product in Figure 3 were as follows:

1. Adjust the CFD model of dimensions in Figure 2 to

match test data within 10% including a mesh sensitivity

study.

2. Make the return air (RA) damper the same size

as the outdoor air (OA) damper, changing it from

54.2 in. × 18.2 in. (1377 mm × 462 mm) to 67.9 in. ×

22.6 in. (1725 mm × 574 mm).

3. Change the box size to that shown in Figure 3

(90.5 in. × 28.375 in. [2299 mm × 721 mm]).

4. Change the RA and the OA damper size to that

shown in Figure 3 (24 in. × 88 in. [610 mm × 2235 mm]).

5. Drop the total ﬂow to 10,000 cfm (4719 L/s) to match

the 25 ton (87.5 kW) rating.

Mixing effectiveness was described in the ASHRAE

Research Report

as both range effectiveness and ther-

mal statistical mixing effectiveness. Robinson

has

deﬁned range effectiveness as:

Range

=−













×1 100%

(1)

where

Range

= (T

max

– T

min

)

Downstream

Range

= (T

max

– T

min

)

Upstream

And statistical effectiveness as:

Statistical

=−













×1 100%

(2)

where

= Temperature standard deviation downstream

of the mixing device

= Temperature standard deviation upstream of

the mixing device

FIGURE 1 Economizer diagram.

Outdoor Air

(OA)

Mixed Air

(MA)

Supply Air

Fan

(SAF)

Cooling

Coil

(CC)

Supply Air

(SA)

Heating Coil

(HC)

Filter

(F)

Recirculated Air (CA)

Dampers (D)

Exhaust Air

(EA)

Return Air Fan (RAF)

Return Air

(RA)

FIGURE 2 Test setup from ASHRAE Research Report RP-1045.

Atmospheric

Pressure

OA Damper

RA Damper

Airﬂow

in.

12 in.

in.

117 in.

38 in.

144 in.

FIGURE 3 Commercially available 20 ton to 25 ton mixing box.

in.

88 in.

24 in.

8 in.

TECHNICAL FEATURE

ASHRAE JOURNAL ashrae.org DECEMBER 202024

As Robinson

points out, each effectiveness method

has its own strengths and weaknesses. Range effective-

ness is good for reporting extremes of temperature that

might lead to coil freezing or nuisance freeze stat trip-

ping. Statistical effectiveness is good at reporting the

uniformity of the ﬂow for control response. However,

in practice, usually a single sensor is placed in each

airstream, which makes statistical effectiveness more a

tool for design and analysis and not relevant for control

strategies.

Getting the standard deviation of upstream tempera-

ture has some inconsistencies in the literature and some

practical difﬁculties in sample size for a standard devia-

tion. Finaish, et al.,

deﬁnes the upstream standard

deviation as:

= 0.5(T

−T

) (3)

where T

is the return air temperature and T

is the

outdoor air temperature.

This deﬁnition differs from the deﬁnition in another

paper, where Sauer, et al.,

deﬁnes the upstream stan-

dard deviation as:

RA OA

−

()

(4)

where V

and V

are the velocity of the return and out-

door airstreams.

This deﬁnition of upstream standard deviation

accounts for different airstream velocities, but was

shown to vary widely when the RA damper area was

changed while holding ﬂow rates and all other vari-

ables constant. The derivations of Equation 3 and

Equation 4 were not provided, and the authors of

this paper found no intuitive way to choose between

them.

A simple conservation of energy evaluation neglect-

ing kinetic and potential energy changes yields this

equation:

R A

+ Q

= Q

(5)

Assuming the density, r

, and speciﬁc heat, c

, change

is negligible, the equation becomes:

= Q

(6)

where Q

, Q

and Q

are the return, outdoor and

mixed air volumetric ﬂow rates, respectively. T

, T

and T

are the return, outdoor and mixed air tempera-

tures, respectively.

Using the relationship between volumetric ﬂows and

simplifying yields the following:

+ Q A

= Q

(7)

= (1 – %OA)Q

(8)

= (%OA)Q

(9)

= (1 – %OA)T

+ %OA × T

(10)

where %OA is the percent of outdoor air.

This is the equation used in an economizer reference

manual

for ﬁnding the mixed air temperature.

Considering T

as the average theoretical tempera-

ture, we can compute a more meaningful upstream

standard deviation for the statistical effectiveness.

iMA

−

()

−

∑

(11)

where the data set consists of T

and T

Equation 11 will be used in Equation 2 for reporting of

statistical effectiveness in this paper because it does not

skew the numbers based on damper sizes as Equation 4

does, and it provides a weighted average of tempera-

tures based on the percent outdoor air, unlike Equation 3.

Computer-Aided Design (CAD) Model

Preparation and Uncertainties

An air mixing box with dampers facing perpendicu-

lar to each other was selected for analysis since this

is the most common conﬁguration. The analysis was

performed in steady state with adiabatic walls. Figure 2

shows the 3D model with dimensions created to match

Conﬁguration 4 tested in Test 7, with the results pub-

lished in ASHRAE Research Report, RP-1045, “Verifying

Mixed Air Damper Temperature Control and Air Mixing

Characteristics.”

To prevent any unrealistic recircula-

tion in the CFD simulation, the inlets and outlet sur-

faces were extended by 100 in. (2540 mm) and 150 in.

(3810 mm), respectively. There is also some uncertainty

on how the real blades were positioned and their open-

ing angles when performing the test.

TECHNICAL FEATURE

DECEMBER 2020 ashrae.org ASHRAE JOURNAL 25

It was assumed that blades opened at 90 degrees

equals a 100% airﬂow through that damper. It was esti-

mated that a 30% airﬂow would occur with the blade

angle at 27 degrees. The exact location of the blades is

also unknown, which might be another factor increas-

ing error between the CFD and the real test. According

to ASHRAE Guideline 16-2018,

the return dampers are

usually low value and will behave more linearly for the

control response as parallel dampers. Also according

to Guideline 16-2018, the outdoor air damper does not

control airﬂow, so it could be parallel or opposed blade.

In this study both dampers are modeled as parallel blade

dampers.

CFD Preparation

A full-cloud computer-aided engineering (CAE)

software with various simulation types was used to

perform the simulations of the mixing box CAD mod-

els. In this case a convective heat transfer simula-

tion was selected with a turbulent ﬂow algorithm. To

select the appropriate turbulent model, the Reynolds

number was calculated assuming the full ﬂow rate of

16,000 cfm (7551 L/s) and is ﬂowing in the direction

shown in Figure 2 with an average air temperature of

90°F (32.2°C).

The Reynolds number was calculated as 244,300,

which deﬁnes a turbulent ﬂow. As recommended by

CFD literature, a K-Omega SST model was selected.

This model is a two-equation model, combining turbu-

lent kinetic energy and dissipation.

Boundary conditions were set as published in the

ASHRAE report

for Conﬁguration 4 and Test 7 using the

normalized temperature rather than test temperatures:

• OA Inlet: 4,800 cfm (2265 L/s) and 30°F (-1.1°C);

• RA Inlet: 11,200 cfm (5286 L/s) and 70°F (21.1°C); and

• Mixed Air Outlet: 1 atm.

As previously mentioned, the range mixing effective-

ness and the statistical mixing effectiveness were used

to validate this CFD simulation to the published test

results. To achieve this, it was required to read the tem-

perature values in the same position as the test probes

in the test report. As shown in Figure 2, a probe grid was

used with 12 in. (305 mm) spacing between each probe,

4.5 in. (114 mm) from the side wall and 7 in. (178 mm)

from top and bottom walls. This produces three rows

of 10 probe points each. Note that these probe points,

which are physically visible in Figure 2 and initially

modeled as features, were replaced with virtual probe

points within the software to reduce mesh size and com-

putational time.

This probe grid is located using coordinates with the

origin in the centroid of the model. As the plenum size

is reduced in the study, the probes remain in the same

location with relation to the center of the model, making

probe points 1, 10, 11, 20, 21 and 30 no longer inside the

control volume being analyzed, as shown in the bottom

portion of Figure 4.

Mesh

The model was run with different mesh settings to test

the sensitivity of the results and how convergence was

affected by the mesh setting. The software allows for the

adjustment of mesh size using a “ﬁneness” value from

one to 10, with 10 being the ﬁnest mesh. Simulations

were run with mesh sizes from six to 10. The simulation

failed with mesh ﬁneness at six. With the ﬁneness at

seven, the simulation was successful and convergence

reasonable, but with a high mixing effectiveness error

compared to the published test report. When reﬁn-

ing the mesh to a ﬁneness of eight, convergence was

improved, and the range and statistical mixing effec-

tiveness values were within 2% and 7% of the test data,

respectively.

Figure 5 shows a cut plot showing the temperature dis-

tribution using a mesh ﬁneness of eight, which yields

range and statistical mixing effectiveness of 61.0% and

83.9% compared to 60.3% and 87.2% of the test data,

respectively. Pressure drop was calculated to be about

0.22 in. w.c. (55 Pa).

FIGURE 4 Probe grid layout for 117 in. × 38 in. and 90 in. × 28 in. plenum.

P1 P2 P3 P4 P5 P6 P7 P8 P9 P10

P11 P12 P13 P14 P15 P16 P17 P18 P19 P20

P21 P22 P23 P24 P25 P26 P27 P28 P29 P30

P12

P22

P19

P29

117 in.×38 in. Plenum

90 in.×28 in. Plenum

TECHNICAL FEATURE