Basic of HVAC Duct Design.pdf

8/10/2019 Basic of HVAC Duct Design.pdf

1/3

P U L L O U T

Basics of Duct Design

This month we begin therst of HVAC&R Nationsthree-part series onducts. Two duct designexperts, JJW (Bill) Sigantoand Murray Mason

provide a wealth ofinformation on the topic.

What is good duct design?Good duct design optimises:

Duct size (manufacturing costs)

Duct systems pressure losses (operation cost)

Duct systems acoustics (environmental costs),

and

Air balance procedures (commissioning costs)

Of these, duct sizing is the simplest. In a few minutes,even the most complex systems can be sized with aduct sizing slide rule. The various texts suggest threemethods:

Constant friction gradient

Constant velocity, and

Static pressure regain

A pragmatic solution to optimise the conflictingelements in duct design, the late W R (Roy) Ahern,offered the following approach (1) for the sizing of

low-pressure air conditioning ductwork.

For air quantities greater than 4500 l/s size ductat 10m/s.

For air quantities less than 180 l/s, size ductat 4.5 m/s.

For air quantities greater than 180 l/s, but lessthan 4500 l/s

Then size duct on basis of friction loss of 1.2 Pa/m

For these rules, Ahern offered no references otherthan his own extensive experience backed by hisunerring ability to reduce all technical programs tofirst principles. In the larger size where velocity is the

criterion, the limiting factor is noise breakout. Hischoice if 1.2 Pa/m as the mid-range f riction gradientis higher than that used by many designers.

For the smaller sizes, the proposed velocity limit

often produces a high fric tion gradient. Consequentdynamic losses are quite low with the resultantvelocity pressure around 12Pa.

Velocity pressure(or velocity head) The terms velocity pressure and velocity head areinterchangeable.

For air at comfort temperature the velocity pressureis 0.6 times the square of the duct velocity.

Pv = 0.6 x v 2, in Pascals (Pa), if v is measured in persecond.

Static pressure frictionStatic pressure friction causes static pressure loss. The calculation of the friction gradient f (Pa/m) iscomplex, involving velocity pressure, duct diameterand duct surface roughness. Modern computers canmake it routine. However, most designers still rely onduct sizing slide rules.

The static pressure loss in a section of duct is:

PT = f x L, where L is the duct length in meters.

Fitting lengths should be included in the estimationof L. Where a fit ting involves a change in velocitythan fitting should be included in the duct sectionwith the higher velocity.

Total pressure Total Pressure is the sum of Static Pressure andVelocity Pressure.

PT = PS + PV

Since one of the variables of friction gradient isvelocity pressure, it follows that total pressure isdependent on the velocity pressure, which for agiven duct area varies as the square of the flow.As air flows through a duct system from a fan toterminal the total pressure is always decreasing.Because of changes in velocity, the same cannot besaid of velocity pressure and static pressure.

FittingsWhen a fitting changes the direction of airf lowor its velocity, then as well as the friction lossreferred to above, a dynamic loss is involved. The quantum of dynamic loss is expressedin total pressure. Various fittings are assigneda fitting loss factor (K). The dynamics lossthrough the fittings is then calculated from,

PT (Pa) = P V

Where the velocity pressures is usually related tothe maximum velocity occurring in the fitting.AIRAH(2), ASHARE(3) and SMACNA(4)are the mostwidely used references for fitting loss factors. Ductdesigners are often frustrated by the disparity

skills workshop

HVAC&R SKILLS WORKSHOP M O D U L E 2 2 P U L L O U T

P U L L O U T

15 Proudly sponsored by

Exposed office ductwork


2/3

skills workshop

16

between the references for K values for the samefitting, particularly for those involving dividedflow. This so concerned Ahern 50 years ago thathe designed his own f ittings and calculatedK values from unassailable first principles.

Long runout of flexible duc t were not extensivelyused in Aherns time, but these days it is reasonablepractice to limit velocity to 3.5m/s and for frictionloss calculation purposes to allow twice themeasured length.

The almost universal use of cushion head involvesan additional pressure loss of at least one and a halfvelocity heads (1).

Balancing designers should ensure that the ductsystem is proper. Unless the designer considers howthe duct sys tems can be properly balanced, morethan likely it wont be. The following is consideredfundamental:

Design for application of proportional balancingmethods

Provide adequate dampers for balance of subsystems

Do not rely on splitter devices as balancing aids

References1 Ahern WR, AIRAH Journal Vol.16, No.6 :Duct sizing,

June 1977

2 AIRAH Application Manual Duct Design DA3, 1987

3 ASHRAEHandbook of Fundamentals 2001

4 SMACNA HVAC Systems Duct Design 3rd Edition 1990.

FurtherconsiderationsDuct sizing methods There is no single duct sizing method that willinherently give the best duct design. Whilst mostpeople are aware of the constant pressure gradient,constant velocity and static regain methods, thereis a further method known as the balanced pressuredrop method and yet another more recent methodknown as the T-Method Optimisation.

The balanced pressure drop method is describedin the AIRAH Application Manual DA3 and involvessizing the duct layout using the constant pressuregradient or static regain method, determining theindex run (the path with the greatest pressure drop)and then reducing the duct sizes in all other paths(without exceeding velocity limits) such that the out

of balance pressure drop in each path is minimised. The objective of this method is to achieve a morenearly balanced system thereby reducing noiseand making the sys tem more easily balanced whencommissioned.

The T-Method Optimisation optimises the ductdesign on the basis of system capital cost and thepresent worth of energy. It is described in detail inthe ASHRAE Fundamentals Handbook.

Fitting lossesWhile fitting losses can be allowed for by allowing anequivalent length, more reliable and comprehensivedata is available in the form of loss coefficients (k).

Care must be taken when using this data, however;because different texts base loss coeff icients ondifferent velocities in the fitting eg. the branchpath pressure loss for a divided flow f itting can be

expressed as a k factor based on the branch duct

velocity or based on the main or upstream ductvelocity. These different loss coef ficients are relatedby:

k u = k B.(VB /Vu)2

Where:

k u = the loss coefficient based on the upstreamvelocity

k B = the loss coefficient based on the branchvelocity

V u = the upstream duct velocity

V B = the branch duct velocity

Hence the pressure loss through the branch path is:

1/2 p x k u x Vu2 = 1/2p x k B x VB

2

(where p = density of air = 1.2 kg/m 2)

Awareness should also be factored in for a numberof fittings, notably bends. The published data mustbe corrected for the angle of turn of the bend. Notethere is also a Reynolds Number correction. This cangive a significant increase in pressure drop at highervelocities (Refer clause 6-30 of AIRAH Applicationmanual DA03)

Fitting interaction

Another important consideration with fitting lossesis that fittings in close proximity can have a higher(and in some cases lower) combined pressure loss.While it is reasonable to say that fittings shouldnot be located close together, particularly in anS configuration, in practice this often cannot beavoided, eg. when ducts have to drop under beams.(Clauses 6-40 to 6-120 of DA3 discuss the effectsof fitting interaction and also the effects of poorlyconfigured fan layouts.)

Duct attenuationPublished data on lined duct attenuation is generallyvery sparse. Much of the data is for only a limited setof sizes making interpolation for intermediate sizesextremely difficult. Duct attenuation is not linear,ie. if you keep increasing the length of duct, theattenuation does not keep increasing in proportion. This is because of self-generated noise in the duct.Suppliers attempt to account for this by publishingattenuation for different lengths of duct. Thus weget the anomalous situation where two lengthsof two metre duct either side of a transition gives(apparently), a higher attenuation to that of a fourmeter length of straight duct.

To determine the attenuation accurately, accountmust be taken of self-generated noise in the duct.

The same applies to fittings. The noise level in a ductsystem does not progressively decrease away fromthe fan until it reaches zero. There is a lower limitcaused by self-generated noise.

Ductwork

Proudly sponsored by

This information first appeared asthe article Basics of Duct Design by

JJW (Bill) Siganto and appeared inEcolibrium magazine.


3/3

P U L L O U T

Next Months Workshop Ducts Part 217

HVAC&R SKILLS WORKSHOP M O D U L E 2 2 P U L L O U T

Self-generated noiseSelf-generated noise is generally proportional tovelocity to the sixth power (pressure is proportionalto the square of velocity), the duct cross sectionalarea, a characteristic dimension in the case offittings and the frequency. The formulas though notexcessively complex, do mean that a complete ductdesign, including an acoustic analysis, is very tedious.It will also necessitate the use of a computer. If noisecriteria of NR 25 or even NR 30 are to be achieved, afull acoustical analysis is essential.

Acoustical analysisof a ductwork system The acoustical analysis of a ductwork system isnormally carried out by calculating the sound powerlevel along the ductwork sys tem, starting withthe fan. Taking account of the attenuation and selfgenerated noise of ducts and fittings, the soundpower at each air terminal is calculated. Some ofthis sound is reflected back up the duct leadingto the terminal. The reflected noise is a functionof the diameter of the duct at the connectionto the terminal (the terminal neck diameter) andthe frequency and design charts are available todetermine this.

The noise coming from the duct (after deducting theend reflection loss) is then added (logarithmically)

to the noise generated by the air terminal. This thengives the total noise entering the room at this point.

The sound pressure level at any listener positionin the room is then calculated from the acousticalproperties of the room (expressed as the roomconstant R), the directionality of the noise from theterminal (expressed as the directivity factor Q) andthe distance (r) from the terminal to the listenerposition. Hence:

Lp = Lw + 10 log Q + 4

4 r2 R

Where

Lp = the sound pressure level at the listenerposition for each terminal

Lw = the sound power level at the terminal

Good duct designAs stated, good duc t design involves sizing theducts, determining the pressure losses, calculatingthe noise levels, determining the out of balancepressures and optimising this against the total costof the system. Whilst systems that are not noisecritical can be simply sized using a duct sizing sliderule, commissioning can be difficult if no thoughtis given to the out of balance pressure alongeach duct path, particularly if the layout is verynon-symmetrical. This can result in unexpectedexcessive noise levels. If the sys tem is noise critical

(eg. a TV studio or theatre complex) then there is nosimple way to design the system and perform theoptimisation manually. Detailed pressure drop andacoustical calculations (including self-generatednoise) must be carried out. s

Duct being wrapped

DA03 Ductworkfor air conditioningAIRAHs DA3 Ductwork for air conditioningprovides detailed guidelines for designingair conditioning ductwork systems. Themanual covers sizing and the completeacoustical analysis of ductwork systems plusdetermination of qualities for accurate design

costing. To order a copy, visit www.airah.org.auor contact AIRAH on 03 8623 3000.

This information first appearedas the article Basics of Duct Design: A further word by Murray Masonand appeared in Ecolibrium magazine.

[ ]

Documents

Basic of HVAC Duct Design.pdf