Designing your dust collection system to meetNFPAstqndqrds r Pqrtl

GaryQ.Johnson Workplace Exposure Solutions

About 40 percent of combustible dust explosions ne-ported in theUS and Europe overthe last25 yeamhave involved dust collectors. Dust collection sys-tems are now a primary focus of inspections re-quircd by OSIIA's National Emphasis Program onsafely handling combustible dusts.l OSHA also hasthe authority to enforce National Fire Protection As-sociation (NFPA) standards for preventing or pro-tecting against dust explosions. This two-part articlefocuses on how you can design your dust collectionsystem's dust collector, ductwork, and exhaust fanto meet the intent of these NFPA requirements. ParttrwillappearinJanuary.

fflhe explosion hazards posed by dusts commonly han-

I dled in bulk solids plants can be surprising. In fact,I most natural or synthetic organic dusts and some

metal dusts can explode under the right conditions. You canfind a limited list of combustible dusts and their explosiondata in the appendix to the NFPA standard focusing on dustexplosion hazards, NFPA 6B: Standard on Explosion Pro-tection by Deflagration Venting (2D7),'� and in Rolf K. Eck-hoff's bookDust Explosions inthe Process Industries.3

While such published data can give you some idea of yourdust's explosion hazards, using this data for designing ex-plosion prevention or protection equipment for your dustisn't recommended. Your processing conditions and yourdust's characteristics - such as its particle size distribu-

tion - differ from those for the published data for thesame material, producing different combustible dust re-sults. The only way to determine your dust's combustibil-ity is to have a qualified laboratory run explosion tests on arepresentative sample of the dust. Then, to meetNFPA re-quirements, you'll need to commission a hazard analysisof your dust collection system to document that its designmitigates the explosion risk posed by your dust. (For moreinformation, see reference 4.)

Some dust explosion basics

The five elements required for a dust explosion canbe pic-tured as a pentagon, as shown in Figure 1. The three ele-ments labeled in black are those in the familiar firetiangle: fuel (combustible dust), an ignition source, and

Dust explosion pentagon

Dust dispersionat or greaterthan dust's MEC

Confinementof dust cloudin equipmentor bui ld ing

(combustibledust)

lgnitions0urce

D E t r E M B E R z û o B I P B E

oxygen. For a dust explosion. two more elements (labeledin red) are required: dust dispersion at or greater than thedust's minimum explosible concentration (the lowest dustconcentration that will propagate a combustible dust defla-gration or explosion; MEC) and confinement of the dustcloudwithin equipment or a building.

Put simply, a dust explosion occurs when an ignitionsource touches a dust cloud with a concentration at orgreater than the dust's MEC. A dust cloud with this con-centration can result when a layer of dust thicker than%zinch on equipment, piping, overhead conduit, or similarcomponents is pushed into the air by some event, such asthe pressure wave from a relief device's operation. Whenan ignition source - such as a spark or the flame frontfrom an equipment explosion - touches the cloud, thedust can explode with devastating impact, as evidenced bythe fatal results of the sugar refinery explosion in Georgialast February. To mitigate your dust collection system'sexplosion risk, you need to focus on preventing dust accu-mulation in the system, preventing ignition, and providingexplosion prevention or protection at the collector - allcovered by NFPA standards.

Even when a dust collector is equipped with an explosionvent that works properly, the ductwork in the dust collec-tion system can propagate a collector dust explosionthroughout a process area. An investigation into one suchcase revealed that a contributing factor was the ignitionand explosion of dust that had accumulated in the duct-workbecause of the system's inadequate conveying veloc-i t y .5 Another cont r ibu t ing fac to r was the lack o ffl ame-front-isolation devices in the collector's dirty-airinlet and the clean-air outlet for recirculating air to thebuilding. Such devices could have prevented the flamefrontin the collectorfrom entering the inlet duct andre-en-tering the building through the outlet duct.

In this case as in many others, following the requirementsin NFPA standards for mitigating explosion risks in a dustcollection system could have prevented the dust explosionfrom propagating beyond the dust collector. In the follow-ing sections, we'll look at how you can design your dustcollection system to meet the NFPA standards. Informa-tion covers preventing dust accumulation in ductwork,eliminating ignition sources, and using explosion preven-tion and protection methods at the collector. fEditor'snote: Captvehood design, another important factor in de-signing a safe dust collection system, is beyond this arti-cle's scope; for more information, see the later section"For further reading" or contact the author.]

Preventing dust accumulation in ductwork

To prevent dust from accumulating in your dust collectionsystem's ductwork and becoming fuel for an explosion,you must design all ducts in the system with two principlesin mind, as described in NFPA 654: Standard for the Pre-

vention of Fire and Dust Explosions from the Manufactur-ing, Processing, and Handling of Combustible ParticulateSolids (2006).'First, the conveying air velocity must beadequate throughout the duct. Second, at points where twoairstreams merge, the duct sections must join in a way thatmaintains this velocity.

If your plant handles a combustible dust, a visiting OSHAinspector will ask whether the conveying air velocitythrough your dust collection system is adequate - andwill ask you to prove it. Why is this velocity important?Keeping the conveying airvelocity in every part of the ductwithin a reasonable range will prevent two problems: Toolow an air velocity will cause the dust to drop out of the airand build up inside the duct, and, depending on the dust'scharacteristics, too high an air velocity will waste energy,erode the duct. or. if the dust is moist or stickv. cause thedust to smear on the duct wall.

lf your plant handles o combustible dust, o visitingOSHA inspector will askwhether the conveying oirvelocily through your dust collection sysfem isnÀan, ,n to - nnÀ ,n t i l l n .L r tn t t la n rn t to i f

A conveying air velocity between 3,500 and 4,000 fpm(17 .5 and20 m/s) is a reasonable starting point for design-ing your system. Then, based on supporting data aboutyour application, you can speed or slow the conveying airto the system's optimal velocity. For instance, if your sys-tem handles an extremely fine, lightweight material thatwon't clump together, like cotton dust, you can slow thevelocity to 3,000 fpm; if you handle a very heavy material,like lead dust, you may need to increase the velocity to4,500 to 5,000 fpm. For guidance in determining the opti-mal velocity for your application, see Table 5-1 in theAmerican Conference of Governmental Industrial Hy-gienists' Industrial Ventilation: A Manual of Recom-mended Practice for Design (26Th edition, 2007),6 whichlists minimum duct desisn airvelocities formanv dusts.

How duct sections are joined in your system also affectsthe conveying air velocity. If incorrectly designed, thepoint where ducts join to merge two airstreams can slowthe air velocity, in turn causing the dust to drop out and ac-cumulate in the duct. You can prevent this problem by con-necting each branch duct to a 1 5 -degree tapered expansionon the main duct, which enlarges the main duct diameterto a size appropriate for the merged airstreams. A relatedproblem is that dust particles can drop out ofthe airstreamwhen a branch duct joins the main duct at too great anangle. The momentum of the conveyed dust particlescauses them to want to move in a straight line, so when oneduct joins another at a sharp angle, the particles have to

change direction and slow down. Avoid this problembydesigning the branch duct entry with no more than a 30-degree angle to the main duct.

Failing to practice these duct design principles can lead toany of several problems that produce a slower-than-re-quired conveying air velocity in your ducts. Following aresome visual clues that indicate the air velocity in the ductsisn'thigh enough to prevent dust from dropping out oftheair. Under each clue, ways to remedy the problem and getadequate conveying air velocity though the ducts are de-scribed.

Clue 1: Main duct diameter doesn't enlarge after branchjunctinns. In Figure 2a, tw o 8-inch-diameter branch ductsjoin an 8-inch-diameter main duct, and the main duct'sdownstream diameter is the same after each iunction. AtA, before the first branch junction. the +.200-fpm air ve-locity required to convey the dust is reasonable and can beachieved by the system's design airflow of 1,500 cfm. Butbecause the duct diameter doesn't enlarge after the branchjunctions, the required air velocity increases exponen-tially: It's 8,400 fpm at B, after the first branch junction,which would require a 3,000-cfm airflow, and it's 16,800fpm at C, after the second junction, which would require a4,500-cfm airflow. However, 5,500 fpm is the practicalupper limit for air velocity in system ductwork. To meetthe velocity requirements in this duct arrangement, thesystem would require a major upgrade of the exhaust fanand electric power, which is impractical.

Solution: The more economical solution is to enlarge thedownstream duct. This will solve the problem that resultsfrom not upgrading the exhaust fan - that is, that C getsmost of the airflow, B gets some, and A gets very little. Toensure that the duct's diameter is large enough after abranch ductjoins it, follow this rule of thumb: The sum ofthe areas of the upstream branch ducts should roughlyequal the area of the downstream duct. Based on the equa-tion duct area = n X (diameter/2)', this rule can be re-stated as: the sum ofthe squares ofthe upstream branchduct diameters should roughly approximate the square ofthe downstream main duct diameter. Thus, at B:

8'+8'= 128- I l ' �or l2I

so the main duct diameter at B should be chansed to 1linches. Then, atC, two solutions arepossible:

I 1 '+ 8'= 185 - 13' �or 169

112+82= 185 - I4 ' �o r196

so the main duct diameter at C should be changed to 13 or14 inches, depending on your application's conveying ve-locity requirements.

Clue 2: Main duct is blankcd ffilnFigtxe 2b, an 8-inch-diameter main duct. A. is blanked off. A 4-inch-diameterbranch duct, B, joins the main duct at a Y junction that en-

P B E I o e c e v B E R 2 o o e 55

larges from 8 to 9 inches, and the downstream main duct,C, is 9 inches in diameter. Before the blank flange was in-stalled, the system's design airflow met the air velocity re-quirements at A (4,200 fpm [1,500-cfm airflow]), B(3,900 fpm [350-cfm airflow]), and C (4,100 fpm [1,850-cfm airflowl). But with A blanked off, the required air ve-locity through the 4-inch-diameter duct (B) is now 3,900fpm (350-cfm airflow). The exhaust fan might be able topull an airflow of no more than 600 cfm through B and C,which would drop the air velocity at C from the required4,100fpmto 270fpm.

Solutions: Two solutions are possible: You can replace allthe duct between B and the system's dust collector withsmaller duct to achieve an adequate conveying velocity.Or, as a much cheaper alternative, you can remove theblank flange and replace it with an orifice plate that deliv-ers 1,500-cfm airflow at the system's available static pres-sure; the orifice plate has a hole at its center that's sized tomeet the system's airflow and pressure drop requirements.

Clue 3: Poor duct junctions don't mnintnin conveying ve-locity.Let's look at two examples of this problem. In thefust, shown in Figure 2c, an 8-inch-diameter duct sectionabruptly joins a 20-inch-diameter section. At the system's1,400-cfm design airflow, the conveying air velocity is4,000 fominthe 8-inch section, butitdrops abruptly to 650fpminthe 20-inch secton, whichwillcause thedustto dropout of the air. Solution: ln this case, the solution is to replacethe 2O-inch duct section with S-inch duct. The duct diametershould stay at 8 inches until the next branch junction; afterthatjunction, the duct shouldbe enlargedto maintainthe airvelocity, following the rule of thumb under Clue 1 .

Another poor duct junction is shown in Figure 2d. Here, an8-inch-diameter branch duct, A, joins an 8-inch-diametermain duct at a 9O-degree angle, forming a T junction. Thesystem's 1,400-cfm design airflow can produce the re-quired 4,000-fpm conveying air velocity through A andand section B upstream from the junction without a prob-lem. But section C downstream from the junction wouldrequire 8,000 fpm (at an airflow of 2,800 cfm) to conveythe dust through the duct and past the T junction. Meetingthis impossibly high air velocity requirement would de-mand an unreasonably high fan energy, and the duct atboth B and C wouldprobably plug with dust. Solution:Inthis case, replacing the Tjunction with a 3O-degree Yjunc-tion that enlarges to a downstream diameter of 11 inches(again following the rule of thumb in Clue 1 ) will maintainthe system's 4,000-fpm conveying air velocity.

Clue 4: Ductwork includes too much flexible hose.InFigure 2e, flexible hose has been used in place of metalduct as a quick way to connect two duct sections. How-ever, dust builds up more easily on the hose's comrgatedinside surface than on smooth metal duct. The hose's inter-nal resistance also is more than twice that of smooth metal

D E E E M B E R Z t r t r E I P B E

duct, so with the hose bends acting as elbows, the hose'sequivalent length is much greater than its actual length.The system's exhaust fan may not be large enough to pro-

vide the speed necessary to overcome this additional air-flow resistance, and the result is low air velocity thatcauses dust to drop out ofthe air and plug the ducts.

a. Main duct diameter doesnl enlarge after branch iunctions

Visual clues to duct problems that cause dust accumulation

Blankflange

4 inches

c. Smaller-diameter duct abruptly ioins larger-diameter duct

30-degreeY junction

Blast gatein unlockedposition

d. Branch ductjoins main ductatTiunction

e. Ductwork includestoo muchflexible hose

b. Main duct is blanked off

f. Duct blast gate isn't locked in position

8 inches

Solution: Replace the flexible hose with sections of metalduct that are clamped together. You should use flexiblehose in the system only with equipment that must move,such as connecting metal duct to the capture hood for aloss-in-weight feeder that rests on load cells; see NFPA654 for more information.

Clue 5: Duct blast gate isn't locked in position. Blastgates in ducts add artifrcial airflow resistance to balancethe airflow in individual duct branches. For each blast gate,only one position is correct to balance the airflow in allbranches. In Figure 2f,the blast gate has been adjusted tosend more airflow into this branch, which steals airflowfrom otherbranches.

Solution: Set this and other duct blast gates to meet the sys-tem's design airflow and then lock the gates in place. Youcan avoid this problem altogether by designing the dustcollection system for the correct airflow balance withoutusing blast gates, which i s called b alanc e by de s i g n.

Clae 6: The pressure drop across the filter media ishigherthanthe designpressure drop. Figure 3 shows theairfl ow resistance increasing in a baghouse dust collectionsystem in which the pressure drop across the bag filtermedia exceeds the design pressure drop. (Pressure drop, ordffirential pressure, is the difference in the static pres-

PBEI oeceveER z t r t r e

sures measured on the clean and dirty sides of the dust col-lector; the more dust collected on the filters, the higher thepressure drop will be.) In Figure 3, the design pressuredrop across the bag filters is the system's design static

Effecton airflowof high pressure dropacrossthe filter media

E=

o " - -

o3q

c

fqq

=.F

(r't

Designstatic pressure

02 olAirflow (in actual cubic feet per minute)

High pressuredrop shiftsfan operationcurve to left

Systemoperatingpoint

Design staticpressureto move airfrom longestduct branch tobaghouse inlet

Exhaustfanoperation

D E t r E M B E R 2 O O B I P B E

pressure (11 inches water column) minus the static pres-sure required to move the air from the longest branch ductto the baghouse inlet (7 inches), which equals 4 incheswater column. The line Qr represents the design airflowthe fan should deliver, with the line's top white portion rep-resentng the design pressure drop though the media. ButQ, represents the airflow the fan actually delivers, whichislower than the design level because the actual pressuredrop through the media (shown by the line's white portion)exceeds the design level, increasing airflow resistance.

The system exhaust fan's operating curve shows howmuch airthe fan canmove (airflow, representedby the hor-izontal axis) at different static pressures (on the verticalaxis). As you can see, this curve shifts to the left of the sys-tem operating point - showing that the fan delivers lessairflow than the system requires - [sç2usg the higherpressure drop has increased the airflow resistance acrossthe system. Because the fan can't deliver the required air-flow, the conveying air velocity in the ductwork slows andleads to more dustdropping outin the ducts.

Solutions: The solution to high pressure drop across themedia depends on your application. Assuming you've se-lected the right air-to-cloth ratio (the airflow in cubic feetper minute divided by the square feet of filter media sur-face area) for your dust collector, properly starting up thecollector when the new filters are installed will provide thebest long-term performance. Once the new filters are in-stalled, you should also condition them (also calledpre-coating or seeding) before your dust collection systemgoes back online; this will build up an initial dust cake onthe media that resists blinding and prevents high pressuredrop. (For more information, see reference 7.) With olderfilters, increasing the cleaning frequency or replacing thefilters more often can control the pressure drop across themedia. Another solution is to replace your filters with onesthat have alarger surface area to better handle your dust-ladenairflow.

Clue 7 : Dust clnud is first sign of trouble. Figure 4 shows adust cloud surrounding a vibrating conveyor that deliverspowderto a sifter; thecloudhas developedbecausethe cap-ture hood over the equipment isn't drawing the dust-ladenair into the dust collection system. The dust cloud - a po-tential explosion hazard - could be the result of duct plug-ging, filter blinding, or other problems, any of which couldreduce airflow throughthe system. Unfortunately, this dustcloud is the first sign of trouble because the dust collectionsystem pressure, airflow, and other data aren't monitored.

Solution: You need to make routine measurements of sta-tic pressure and airflow at appropriate points in the dustcollection system, as well as measure each capture hood'sface velocity (the air velocity at the inlet opening). Suchsystem monitoring will reveal any changes in pressure, air-flow, or face velocity from the system's baseline perfor-

Lack of system monitoring makes dustcloud first sign of trouble

mance data. By helping you spot such changes early, mon-itoring allows you to catch a small problem before it cancreate ahazardous dust cloud in your process area.

What is baseline performance data? lt's documentedproof of your dust collection system's performance atstartup (or after any significant system rnodification),which demonstrates that the system can deliver the designairflow at every capture hood or other dust-controlledopening. This is one of the requirements of NFPA 9l:Standardfor Exhaust Systems for Air Corweying of Va-pors, Gases, Mists, and Noncombustible ParticulateSolids (2004)', which is incorporated by reference intoNFPA 654.lnaddition to verifying the system design, thebaseline data provides a reference point for system moni-toring. Baseline data documentation is powerful evidenceto show an OSHA inspector that your system has adequateconveying velocities. The design documentation youshould keep on file includes the system schematic, a tablelisting locations and dimensions of air-balancing devices(such as blast gates), the as-built system's static pressurebalance calculations for sizing the exhaust fan, and the de-sign bases and specifications for system equipment. (Formore information, see reference 8.) Turning the baselineperformance data over to the operators once the system isonline allows them to use the data to monitor system per-formance and keep the systemworking overthe long term.

In some cases, comparing this baseline data to current op-erating data may help you determine that the original sys-tem design can no longer handle your application'schanged field conditions and requirements. In this case,you'll have to redesign the system to meet the new require-ments. Several situations requiring you to test the dust col-lection system to demonstrate that it works as designed arelistedinNFPA9l. PBE

Nert month: In Part II, sections will cover how to eliminateignition sources in the system and how to use explosion pre-

vention and protection methods at the collector. A final sec-tion will explain how to meet additional NFPA requirements.

Referencesl.The OSHA National Emphasis Program (NEP) directive on safely

handling combustible dusts is uuuilubl" at http://www.osha.gov/O shDoc/Direcûve_pdf/CPl_03 -00-00 8.pdf .

2. Available from National Fire Protection Association (NFPA), IBatterymarch Park, Quincy, MA 02169-7471; 800-244-3555, fax617 -77 0-07 00 (wwwnfpa.org).

3. Rolf K. Eckhoff, Dust Explosions in the Process Industries, ElsevierPublishing, third edition, 2003; available from the PBE Bookstore atwww.powderbulk.com.

4. Lee Morgan and Terry Supine, "Five ways the new explosion ventingrequirements for dust collectors affect you," Powder and BulkEngineering, July 2008, pages 42-49; see "For further reading" forinformation on purchasing a copy of this article.

5. Investigation Report: Combustible Dust Hazard Study, Report 2006-H-1, US Chemical Safety and Hazard Investigation Board,November 2006 (www.chemsafety.gov).

6. Available from American Conference of Governmental IndustrialHygienists, 1330 Kemper Meadow Drive, Cincinnati, OH 45240;5 l3-7 42-6163, fax 513-1 42-3355 (wwwacgih.org, mail @acgih.org).

7. Ed Ravert, "Precoating new filters for better airflow, longer filterlifei' Powder and BuIk Engineering, October 2006, pages 21-33; see"For further reading" for information on purchasing a copy of thisarticle.

8.Gury Q. Johnson, "Dust control system design: Allowing for yourrange of process conditions and establishing baseline performance

PBEI oeceMBER z t r t rB 59

values," Powder and Bulk Engineering, October 2005, pages 39-48; see"For further reading" for information on purchasing a copy of thisarticle.

For further reading

Find more information on designing dust collection sys-tems and preventing dust explosions in articles listedunder "Dust collection and dust control" and "Safety" inPowder and Bulk Engineering's comprehensive articleindex (later in this issue and at PBE's Web site, www.powderbulk.com) and in books available on the Web siteatthe PBE Bookstore. You can also purchase copies ofpast PBE articles at www.powderbulk.com.

Gary Q. Johnson is principal consultantwithWorkplaceExp o s ure S o luti o n s, 7 I 7 2 Will ow o o d D riv e, C inc innat i,O H 4 5 24 1 ; phone/fax 5 I 3 -777 -4626 ( gary @ workexposoln.com, www.workexposoln.com). He holds a master'sdegree in business administrationfrom the University ofScranlon, Scranton, Pa., andabachelor's degree inchemi-cal engineeringfrom Ohio State University in Columbus.