Fan Matrix White Paper

In a previous post I discussed the advantages of the Climate Craft Matrix fan array system.

Climate Craft has now published a white paper that explores these advantages in more depth.



This paper discusses applications and advantages of this system, including:

• VFD Considerations and Electrical Requirements
• Sound and Efficiency Considerations
• Vibration
• Space Considerations
• Reliability
• Serviceability
• Fan Isolation (Backdraft Dampers)
• Cost
• Common Options

This white paper is well worth review and will help designers and owners evaluate the best applications and advantages of this fan innovation.

Fan Engineering: Spark Resistance Ratings

Every once in a while we will see a specification for "explosion proof" fans. While this may be a desirable characteristic, "explosion proof" is not a specifiable option, and usually is included because of confusion with electrical component (i.e. motors, disconnects) specifications.



Instead, fans are generally classified by "spark resistance". AMCA has created a standard that defines three different levels of spark resistance, classes A, B and C.

These classes, listed in decreasing order of assurance, are generally concerned with the prevention of sparks caused by the rubbing together of spark-producing metallic components (generally ferrous materials). These classes only address spark risks due to an explosive airstream, and do not address explosive conditions outside the fan. A summary of the different levels of protection is found in this helpful engineering paper from Twin City Fans.

Type C: The fan is designed so that if the impeller or shaft comes loose and shifts during operation, two ferrous parts will not come into contact.

Type B: In general, this requires a nonferrous impeller and a nonferrous rubbing ring around the shaft hole. Also, extra locking systems are required to prevent the fan impeller, shaft, and bearings from shifting.

Type A: This requires a nonferrous airstream. Also, the extra locking systems are required as in Type B.

As with any engineering decision, the correct level of spark resistance to specify depends strongly upon the particulars of the project: The gasses or substances expected, the concentration of these contaminants, the location of the air-moving device, etc.

Direct Drive, Evolved

Previous articles on this site have discussed the advantages of direct-drive plug fans and the technical tricks required to apply them correctly. However, despite their many advantages, there are times that direct-drive fans just haven't made sense.

In large part, this is because direct-drive fans have been applied as if they were belt-drive fans. It turns out, however, that there is a better way to apply these fans.











See, the problem with direct-drive is that due to the peculiarities of motor performance (discussed in the links above) you usually want to select your fan at a design speed very close to a synchronous motor speed (900, 1200, 1800 rpm, nominally). This limitation can be made up for by varying the width of the fan wheel, but this can cause an unacceptable decrease in static efficiency, or an unacceptable increase in fan noise. Or it can lead to the use of an oversized, less readily available low-rpm motor.

Another strategy is to consider selections of multiple fans, which opens up more design possibilities. However, in standard HVAC designs, this option has practical limits in the number of fans that can be arranged in a cabinet. In the traditional belt-drive paradigm, one or two large fans are mounted on the air handler floor. In unusual situations, three or more fans can be arranged this way, but this requires unusual cabinet geometries that are not often appropriate. This limits the number of direct-drive solutions that can be brought to bear, limitations that are not present with the infinitely-variable fan speeds that are available with belt-drive equipment.

But with a deceptively simple re-thinking of a traditional fan mounting, it becomes possible to stack fans one above another in an air handler cabinet--and suddenly a whole new universe of design solutions present themselves.

It is this evolved fan mounting that is the basis of the ClimateCraft Matrix system.

Matrix is an array of direct-drive plug fans designed to allow maximum flexibility in the selection of fan performance to maximize the benefits of direct drive without the traditional tradeoffs that used to be required. Five fan wheels between 16 and 27 inches are available, with motor sizes between 3 and 30 hp. The fan wheels are AMCA-certified welded aluminum wheels. The wheels are ‘modified class II’ to cover up to 11” static, or class III for higher pressures. The motors are premium efficiency, VFD compatible, 1600V insulation, ODP or TEFC—off the shelf replaceable.

An obvious temptation when mounting multiple, small fans is to avoid the costs associated with isolation and to mount them rigidly to the air handler itself. This simplistic approach, however, can result in repercussions downstream. In fact, the ASHRAE Applications handbook, chapter 47 recommends spring isolators on fans operating above 500 rpm with brake HP below 40. The issue isn’t so much transmitted vibrations, although that certainly can be a problem, but instead bearing life. Strong vibrations can kill bearings, and when the fan bearing is also a motor bearing as is the case in direct-drive plug fans, a bearing failure can be awfully expensive. Climate Craft avoids this problem by isolating every fan from the air handler structure with a unique three-point, seismically-restrained spring isolation system to prevent developing harmonics and creating damaging vibrations. But they have taken the effort even further and used a finite-element analysis to ensure that no harmonic frequencies exist in their fan base anywhere in the operating RPM of their fan systems. This step essentially converts every individual fan base into an inertia base.

This measure ensures that no VFD frequency lock-outs are needed to prevent violent vibration at the fan—a step that is often overlooked in commissioning and can cause unacceptable rates of motor failure. The unique isolator design has the added benefit of preventing the fan base from contacting the seismic restraints and causing a short-circuiting of the fan vibration directly to the frame of the unit and thus the building. This sort of grounding out of the seismic isolation is common on variable airflow systems where the fan thrust changes depending on the fan speed required at any given service point.

The Matrix allows fans to be stacked in towers 1, 2 or three high inside an air handler cabinet. Multiple towers can be installed across the air handler air tunnel. Since the fan wheels are much smaller than typical for the air handler size, many such towers can fit horizontally where only one or two typically sized fans could fit before. These fan towers are designed so that only two different designs are required to support all five different fan wheels and any required motor frame. This greatly aids in construction, making this approach a very cost competitive approach to fan mounting. The towers also serve as a rigid support truss for the fan inlet wall to support the interior of the unit at a location where the pressure differences are often quite high.

Operator Benefits

This sort of a change in fan concept represents a significant advantage for the operator of these systems. This system enhances the air handler’s reliability and serviceability. Redundancy is almost total, since in multiple-fan arrays, the loss of a single fan can often be overcome by the remaining fans simply by ramping up the RPM slightly. Replacement of a failed motor is also much easier. First of all, Matrix uses off-the shelf motor sizes that are easily obtainable on short notice. So simply getting a replacement is easier. Additionally, a typical Matrix motor and wheel assembly might weigh 150 lbs and be easily maneuvered into place by a couple of men, while a typical large fan motor may weigh 1500 lbs or more, and require special rigging to get into place. This may require a significant facility shutdown or even crane work in some cases.


Additionally, the multiple fan array allows shorter air handlers, making a more efficient use of valuable facility square footage. The smaller, faster fans also shift the acoustical signature of the system into higher octave bands, making sound attenuation easier and less expensive.

And a maintenance person will never have to tighten or align a belt on a Matrix system.

Matrix is the next evolution in fan system design.

So, Why Use Direct-Drive, Anyway?

Why indeed?

If you have read my posts on the Hollisterian and Florentine effects on direct-drive fans, you may be wondering if it is really worth the complication to chose this type of fan-drive system for your project.

But when you think about it, the lesson from those two effects is to strive to use fan selections that are at a synchronous speed when at design. And since you know you can vary the width of the wheel to get the CFM you need, this should be a simple trick that you (or any decent AHU provider) can do when laying out your equipment. And if this isn't possible, you know how to compensate for an asynchronous fan selection in your motor size and system design. The only question left is what do you gain for this effort?

Plenty.

Efficiency for one. Belts are a source of inefficiency. Friction between the belt and the pulleys causes heat and erosion of the belt which is exhibited in the system by a reduction in efficiency of the system. How much friction are we talking about? A sample chart might help quantify this:



That's right--About 4% of your motor energy is lost on systems with brake HP's around 50, and more than seven percent on systems less than two BHP. That's a lot of energy to just throw away.

Another reason is maintainability.

Guess how much belt maintenance needs to be done on a direct-drive system? Guess how many belts need to be stocked to replace broken belts? Guess how much time needs to be spent adjusting belt tension to spec? Guess how many times someone needs to be called in to correct a squealing belt?

A rule of maintenance is that if something is hard to do, it won't be done. A good corollary to that might be that if something doesn't need to be done at all, there's a good chance it won't cause a problem downstream because someone didn't do it.

A related advantage is the longevity of the system. For starters, A belt drive system requires at least four bearings, two at the motor and two at the fan. Since a direct drive fan only has the motor bearings, simple mathematics would indicate that you would have at least half the bearing failures. But, as is often the case, the real situation is a bit more complicated than simple math. In each system there is a force on the motor bearing perpendicular to the shaft. In the belt drive system, this force is from the belt tension, in the direct drive system it is from the fan wheel weight.

This is where real life makes things more complicated--because the belt tension is usually several times the weight of the fan wheel. This means that the bearings in the direct-drive fan motor see much lower stress than in the belted case. This translates to many times the expected life for these critical components.

Another advantage? No belt dust. This means that projects with critical air quality concerns may be able to avoid final filters

And let's not forget belt noise--Not just the squeaking that is caused by the slipping of an incorrectly tensioned or worn belt, but the inherent noise that is added to the system from the normal operation of the belt drive itself. (Which is not, by the way, accounted for in the fan sound data you get from the manufacturer. Those were measured on direct-drive fan wheels....)

So let's review:


(Some people might argue that changing pulleys to adjust speed is less of an advantage than a disadvantage)

Now, of course, there will always be applications for belt-drive fans. Sometimes they just make a better fit for the project than direct-drive. Since the motor on a belt-drive fan is supposed to be operating at its synchronous speed (by design) you don't need to oversize motors to reach operating points where a full wheel width is truly the most efficient or quietest solution possible. But direct-drive sure makes sense when it makes sense.

The Florentine Effect, Explained

This post is a follow-on to my earlier posting on the "Hollisterian" effect and is a further explanation of the tricks involved in properly selecting direct-drive fans in air handlers.

By now, you should be familiar with the Hollisterian reduction in available horsepower that occurs when you select a design point at an asynchronous motor speed. And you know that if you ever do need to design such a case, you will also need to oversize your motor by a factor equal to the motor design RPM divided by the actual operating RPM. Thus, if you are laying out an 1800 RPM motor to operate at design at 1500 RPM (let's ignore the fact an 1800 RPM motor is actually 1775 RPM, for now), you will need to select a motor nameplate HP that exceeds the brake HP of the design point by a factor of 1.2 (1800/1500). And because you are such a careful engineer, you've thrown in a bypass around the VFD in order to allow fan operation even if the VFD burns out.

So you are ready to go, right? What else could possibly trip you up now?

This is where the Florentine effect can get you.

The Florentine Effect

Let's go back to that 13.5 HP selection at 1500 RPM we discussed in the last post. Turns out, that's pretty close to a 77% width 30" fan operating at 14,000 cfm and 4" of static. When I select an EPQN fan using the Twin City Fan software, I get a brake horsepower of 13.22 HP. You know that in order to account for the derate due to the RPM that you need to pick a motor that can provide a nominal HP at least 1.2X this brake, or 15.9 BHP. You select the next larger size, or the 20 HP motor. That's a full 50% bigger than the design brake, and at least 25% bigger than you need, when you account for Hollisterian effects.

So you install the fan, and everything works just fine. In fact, you might go several years before any trouble raises its head. But then, suddenly, one day it does.

Let's say the VFD serving the fan burns out, or is taken out of service for a short while. The owner, wanting to preserve function of his fan system, even if he has to operate it at constant speed, does the obvious thing and flips the fan to bypass...

Suddenly the fan kicks on, starts roaring, and then the motor burns out. A follow up inspection might even find that some of the duct fittings have blown apart. What happened?

Well, let's think about what happened. The fan normally operated at 1500 CFM at peak design. The bypass, which is essentially a standard motor starter wired in parallel to the VFD, kicked the motor on at the line power frequency of 60 HZ, or 1800 rpm. This means that you weren't supplying air at 14,000 CFM at 4", but something greater. Following the fan laws, you would ride the system curve up to about 17,000 CFM at 5.5"! And that is assuming that you are operating a constant volume system or a VAV system at peak cooling load. If you were in heating on a VAV system with the boxes choked back to their minimum flows, you might develop even higher pressures!

Let's look at what happened:

(click to see larger image)

The above is a fan curve plot of the two operating conditions of your fan: 1500 RPM and 1800 RPM. The HP curves are plotted for both cases also. They share a common system curve (this assumes an unchanging duct system--a bad assumption for a VAV system). I've highlighted the resulting flow rates and brake horsepowers at the two resultant operating points where the system curve intersects the fan curve.

The first thing that should jump out is that the brake horsepower for this fan operating at 1800 RPM jumps up to 22.9 HP*--even greater than the 20 HP that you picked to protect from the Hollisterian effect! The other thing you should notice is that if this is, in fact, a variable volume system, the system curve shown at design is not the system curve that would be seen by the fan unless the system was at full cooling. If the boxes neck back at part load or in a heating condition, the system curve shifts to the left, pushing the intersection between the fan curve and the system curve closer to the fan curve peak. This greatly increases the amount of static pressure the fan can develop at this higher speed--up to about 9" in the case shown here.

*this can be calculated by the fan law formula:

bhp₂=bhp₁(rpm₂/rpm₁)³

Thus in a VAV system, this can be a double whammy, kicking out your motor and damaging your duct system.

How to avoid this problem? Well, you could size the motor for the even larger size demanded by the Florentine effect--but that would still leave you with the possible problem of overpressurization of the ductwork in bypass. Probably the first thing to consider is whether or not you really need a bypass, anyway. With today's more reliable VFD's, putting in a bypass is far less of a necessity. Many drives can function for the life of the equipment with no failures at all. If redundancy is absolutely necessary, consider providing a second, parallel VFD instead of a standard starter. VFD prices have come down considerably since the parallel-starter bypass concept was developed. This is no longer the cost-prohibitive strategy that it was at one time.

The Hollisterian Effect, Explained

One of the best things about working with seasoned experts is that you get to benefit from their previous, um, experiences. You don't always have to learn the hard way yourself.

Sometimes, these not-quite-the-way-I-planned it episodes are actually elegant illustrations of physical principles, and deserve something more fitting than being remembered as that one time someone screwed something up. Two particular examples certainly fit this bill, and, as it turns out, they both have to do with applying direct drive in custom air handlers. The principles that they illustrate have been christened the "Hollisterian" and "Florentine" effects by our own Jake Marley, in honor of certain colleagues who shall remain unidentified for the purposes of this post. I will discuss the Hollisterian effect here, and the Florentine effect in a future post.

And, instead of dredging up the actual events that gave rise to the discovery of these principles, the gist of which I am sure most readers could figure out, I will instead focus on the principles themselves.

The Hollisterian Effect

Direct Drive fans offer some great advantages to a system designer. There are no belts to maintain, no belt dust to foul the discharge air, no inefficiencies from the belt drive and far less vibration than a belted system. But they also do carry some design limitations that must be dealt with appropriately.

The first limitation? Direct drive fans are direct drive. In other words, they are directly coupled to the motor shaft, and therefore turn at the speed of the motor. Which is great, if you have a design condition where the fan needs to turn at 1800 or 1200 or 900 rpm. If you have a design condition that requires a fan selection at, say, 1500 RPM, then you need to do pick a motor/fan system at an 'asynchronous' design condition.

No big deal, right? We've got VFD's today, so this is a piece of cake.

This is exactly where the Hollisterian effect can get you. See, VFD's are not constant horsepower devices. They are, up to 60 Hz, constant torque devices.

Let's look at the equation for motor power:

hp = (Torque x Speed)/5250

If you have a constant-torque motor, this equation simplifies to"

hp=C x Speed

Where C is a constant equal to the torque constant divided by 5250.

So, what you have got is something like this:

(click on image for larger view)

Where the HP available (the blue line) increases linearly up to 60 Hz, at which point the HP then remains constant and the available torque drops away.

So what does this mean to a designer?

Well, let's say you selected a direct-drive fan to meet your design criteria at a 1500 rpm design condition. Let's say the brake horsepower of that fan selection is 13.5 HP. You select a 15 HP, 1800 RPM motor driven by a VFD. You're good, right?

Well, let's look back at our HP equation--Applying the math, you now have only 1500/1800 (or 5/6th) of the motor hp available at 1500 RPM, or, in this case, 11.7 HP. You really needed a 20 HP motor!

If you are working with low speed fans, and you are selecting in the 400-500 RPM range, you can see that you are going to be robbing about half of the nameplate HP from the selected motor, assuming you are going to select a reasonably available standard motor speed. In the above example, that would turn the 20 HP motor into a 30 HP motor!

So what do you do? Well, one way to attack this problem is to select the bigger motor (and VFD) and call it good. Other than some additional first costs, this might be the right solution. A more elegant solution might be to see if you can't select a fan wheel with slightly shorter blades to bring your design condition in closer to a synchronous speed. Energy Labs provides direct drive systems regularly, and thus will allow you to select plug fans from 50% to 105% of the standard AMCA wheel width to address these sorts of issues. Aaon's fan selection routine in their Ecat32 software allows for variable width wheels and actually takes into account any Hollisterian or Florentine effects (discussed later) in the sizing of their motors!

Variable-width wheel selections allow you to shift the whole fan curve leftwards on the page without losing height--reducing CFM to match your needs, but preserving peak static pressure. This means you could select a fan wheel at 1800 RPM, but reduce the total air delivered by providing a 80% wheel width so that you don't exceed your design flow at the faster speed.

Or, lastly, you could instead chose a 1200 RPM motor, and just select it for 72 Hz service. As long as the motor and drive manufacturer are happy with this selection, there is nothing preventing you from over-speeding your motor.