Combustion Design

[finiteparts]

In thinking how to restart myself on this thread, I thought it might be good to just retouch on the basics of what we are trying to do with each step of the design that has been discussed thus far.

At the most basic level, what are we trying to do here?

The first thing we do is to size the flow restriction through the liner to provide a sufficient level of turbulent mixing in the combustor.  This minimum level of turbulent mixing is required to make sure that the fuel and air are uniformly mixed at the molecular level, where all chemical reactions take place.  The size of the flow restriction is set by the total effective area of the holes and leakages that pass from the compressor discharge to the turbine inlet.  Hopefully, this is almost exclusively done through the drilled holes we put in the liner, since leakages contribute nothing to aid in the combustion efficiency of the burner.

Once we have set the through flow area required to achieve a certain level of pressure drop, we need to figure out how to distribute this airflow.  The bulk of this work is tied to getting the primary zone airflow correct due to the pivotal role that the PZ plays in setting the stability and efficiency of the combustor.  The initial step is to get the fuel to air ratio (FAR) at the design point correctly set in the PZ by sizing all the airflows that enter to achieve the required mass balance.  The important thing to realize at this step is that your primary goal is to set the correct chemical reactor conditions in the PZ to achieve efficient combustion. The local FARs set the airflow partitioning between the primary, secondary and dilutions zones.

With the airflow distributions set, we can go about the task of sizing the passages and the orifices through the liner to meet these target airflow distributions.  The direction and speed of the flow in the outer annulus can influence the penetration depth and the jet angle as it enters the combustor.  Both of these features have dramatic impacts on the quality of the flowfield inside the combustor and thus, the quality of the combustor exit flow that enters the turbine.  Losses in the passages do nothing to aid in the levels of mixing in the combustor and as such, every effort should be made to minimize these losses.  The selection of an accurate hole discharge coefficient is paramount to achieving the correct airflow at each station of holes.  I would also include in this step that in order to properly partition the PZ airflow, the flame stabilization mechanism must also have been selected.

I will try to touch back on each of these points to help clarify them as well a few other things that are good to think about, such as hot to cold geometry, fuel and ignition placement, etc..  Hopefully, I can give the steps a little more clarity and finish up the sizing example so that others can work through this themselves with

a better understanding of what they are trying to achieve.

- Chris

[finiteparts]

If we expand on the first step a bit, hopefully, it will become clear why this has such a critical impact on the efficiency of the combustor.

So earlier in this thread, I worked through the Lefebvre's reference conditions, the pressure loss factor and finally solved for the required effective area needed to meet the selected 5% pressure drop.  I touched briefly on the work by Jim Swithenbank in the mid 1970's to mathematically derive the basis for a minimum required pressure drop based on the dissipation of the mixing energy at the smallest vortex level (Kolmogorov scale).  This is important because you have to remember that to react the fuel and air, they have to be mixed at the molecular level.  So if you mix down to the level where the turbulent energy is finally dissipated into heat, that is the lowest level that you can still stir the fuel and air together.  This is the most mixed that we can do with our turbulent mixing energy, thus increasing the mixing energy further (higher pressure drop) begins to get diminishing returns. Swithenback's work shows that this is in the 3.5 to 5% dP/P range for typical combustor Mach numbers that we might see. 

A side note...remember that dP/P is the change in pressure across the liner over the inlet pressure:

    dP/P = (Inlet total pressure - combustor static pressure) / Inlet total pressure

I also failed to discuss at the time that this dP/P target is usually at the design point, which for our turbo based engines is usually the highest operating point.  This is an important point because like a compressor map, the combustor pressure drop changes as the corrected mass flow varies through the system.

As can be seen, at lower operating conditions, the lower dP/P (which can be taken as the available mixing energy) decreases the combustion efficiency.  At the extreme, the lower combustor efficiency can mean that even with ever increasing fuel addition, there is insufficient fired torque to overcome the compressors demand and thus the engine cannot start.  At a more general case, the lower combustor efficiency can lead to delayed combustion and bits of flame extending past the combustor exit plane.

Well, we are getting some thunderstorms coming through, so I will end this discussion for now...but stay tuned, there is more to follow.

- Chris

[lofi]

Really enjoying this thread, thank you for taking the time!

[finiteparts]

Thanks! I am glad to see the positive response to this thread, I will try to do my best to keep it interesting and relevant to our hobby.

So just a few final thoughts on this first step, which is to size the flow restriction through the liner.

The effective areas of orifices connected in a parallel manner can be added together to produce an equivalent overall effective area.  This can actually be arrived at just by your typical experiences of flows in your day to day life.  If you have a bucket with a hole in it and you pop a second similar hole in it, you would expect that the flow rate leaking out of the bucket to be twice the original leakage.   So we can think of the liner through-flow area as a single orifice for the purposes of this discussion.

If you have an orifice with a fixed area and pressure ratio across it, there is a fixed flow that will pass through it.  Changing the pressure ratio across the orifice changes the flow rate capable of passing through the orifice.  This can be seen in the shape of the map shown in the last post and it can be seen to be starting to "flatten" as it climbs to a choking pressure ratio.  Of course it will not completely flatten out as the orifice throat chokes, since as the upstream pressure increases, the density changes at the orifice and amount of gas that can pass through the orifice at the sonic velocity also increases...but for combustors, we never get anywhere near those Mach numbers.

In reality, the single orifice area model will be wrong because the discharge coefficients (Cd) at each of the different orifices in the liner change as the through-flow Mach number changes.  It is typical to see the Cd's increase with a higher mass flows (which can also be interpreted as higher liner pressure drop or jet Mach numbers) so the curve would likely flatten out faster than with the single orifice model, but for the purpose of hobbyist estimating their own combustor flow-map, this is more than adequate fidelity.  

So now, hopefully it is very clear to the reader that if we start from a fixed mass flow and the initial calculated effective area target, and then somehow experience a larger effective area perhaps due to extra leakage gaps, drilling a few extra holes or even under-predicting the Cd for the hole, the pressure drop across the liner will decrease.  This decrease reduces the mixing energy in the combustor PZ and has the potential to reduce the combustion efficiency or stability.   

Conversely, what if we get the through-low area too small?  Well the concern could be that the extra pressure loss in the cycle, coupled with the low efficiencies of turbocharger components, could lead to such a low overall cycle efficiency that the engine would run hot or not at all. As with all engineering, there is a trade to be made here.

Now, I haven't worked out what sort of effective area shift would cause a significant impact in the dP/P, but my guess is that it would have to be relatively large to cause any significant combustor efficiency or stability issues.  So I am not trying to scare anyone into thinking that these areas have to be spot on, but I do want to impart to the reader the understanding how changes impact the overall system.

- Chris

[racket]

Hi Chris

Would you like to do a worked example for us as its often easier for us DIY'ers to understand than a theoretical non dimensional design.

I used journals.sagepub.com/doi/abs/10.1243/PIME_CONF_1968_183_245_02 when designing the TV84 combustor , not that I knew what I was doing at the time , but it'd be interesting to see what mistakes I made , I have a fair bit of info on the outcomes that might be able to be cross referenced



1.8 lbs/sec at 3.8:1 PR at 70% calculated/measured comp effic

Flametube 145 mm dia by 400 mm long , total volume ~ 1/4 cubic foot

Swirl vane passageways were 9 X 5 mm X 12 of

Cap had 36 X 3mm dia plain holes

Primary 12 X 8 mm plain holes

Secondary 15 X 11mm dia plunged/bellmouthed

Tertiary 10 X 16.5 mm plunged/bellmouthed

The wall cooling louvres were equivalent to 23 X 3 mm holes at each of the 3 bands .

It ended up with ~29% primary 26% secondary and 45% tertiary .

At a latter date I added 23 X 5 mm dia holes at the bottom of the Primary Zone to help the louvres cool the wall as there was heat damage to the 0.5mm stainless on the louvres where they projected into the flametube.

TIT ~900 C

Jetpipe total pressure ~12 psig

Cheers
John

[finiteparts]

I thought that I might skip ahead to the third step, where we are sizing the outer annulus height and the liner holes based on the required effective area and data on the discharge coefficients that is published in various reports. I will return to the partitioning of the airflow between zones after I cover this step, which seemed more suited for discussion after the previous posts.

The penetration and jet angle of air jet that forms in the combustion liner is controlled by the upstream conditions (in the annulus) and the downstream conditions (inside the liner).  If we look at the plain hole through a thin liner we can see that the jet tends to lean in the same direction as the bulk flow momentum that is in the annulus.



The plain hole can be seen to have a very small length to diameter ratio due to the thin sheet that makes up the liner.  This means that the hole cannot impose much flow control on the jet and thus the jet does not turn a full 90 degrees.  It can also be seen that the way the streamlines catch the edges of the hole causes the formation of a narrower "vena contracta" downstream of the hole.  The reduce flow area of the vena contracta limits the mass flow through the orifice and thus acts to reduce the Cd.

Remember that the definition of the discharge coefficient, Cd, is:  

         Cd = actual mass flow / ideal mass flow

So a larger Cd means that there will be a larger flow through to hole for a given pressure loss across the liner.  And just in case there are readers that didn't see the information elsewhere on this forum, the effective area is defined as:

       Effective area = Cd * Geometric Area

You may have anticipated the fact that since the plain hole cannot impose any real directional control on the discharged jet, this might cause problems when the annulus flow direction is not well controlled.  Ideally, the annulus flow has no momentum effects on the local jet as if it is being fed from an upstream plenum, but in reality, the annulus velocity is far from zero.  This is the reason that you need to be careful with how the compressor discharge flow enters the combustor case.  Flow recirculations and directional changes in the annulus flow can cause very distorted reaction zones in the liner.  The downstream effects of these distorted reaction zones can be a very uneven turbine entrance flow.

I have been working recently to design a more uniform entrance flow to a can style combustor and it has become increasing apparent that there is no magic bullet.  Tangential entry can provide a more uniform feed to the liner holes, but it also comes at the price of a slightly reduced feed pressure.  Feeding the combustor from the dilution end, with the airflow travelling forward to the PZ causes the jets to enter the liner with a forward lean.  For the dilution jets, this can be ok, but for the primary jets, this behavior acts to reduce and destabilize the recirculation bubble.  The side entry design causes a stagnation region to occur 180 degrees from the flow entrance, which means that each jet entering the combustor around the perimeter, have a different feed pressure and direction of entry.  I was a fan of most of these designs, but significant CFD work has led me to really rethink how I will feed my next combustor design.  Ironically, the tangential entry and the long pipe feeding the combustor with a well shaped diffuser from the PZ end seem to provide the most uniform feed.

When the plain jet fails us, we always have the ability to increase the holes L/d ratio by making plunged or chuted holes.  Here is the plunged hole schematic:



As can be seen in the schematic, the limiting streamlines impact much farther into the turn and thus the plunged hole can impose a larger control of the jet discharge angle.  Also, as supported by Cd data, the flow control helps to reduce the losses in the jet formation and thus the vena contracta is less pronounced.  If, there happened to be a region of very high annulus momentum that would otherwise cause very scewed jet deflections, even more flow control can be achieved by extending the plunged region further into the liner, producing a chuted hole.  Rolls Royce and Turbomecca both use these effectively in various designs to put airflow exactly where they need it.

There is a great reference out there that is freely available for standard holes on the NASA Technical Report Server.  If you search for TN 3663, you will find the report

ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19930084890.pdf

"Discharge coefficients for combustor-liner air entry holes I - Circular Holes with Parallel Flow". 

It has plots of plain holes in the diameters that we might use, ready to go.  If you dig through the report, you will stumble on this plot:



The primary point of interest in this plot is to show how the CD is reduced as the crossflow velocity is increased.  One of the things that I will touch on in the future is that the airflow velocity on the annulus side of the liner helps to carry away some of the heat that is being imposed onto the inner liner surface by the flame and hot gas, so there is a demand on the heat transfer design side of the combustor to maximize the annulus flow velocity.  But we know that the annulus velocity hurts the ability to feed the liner jets and also increases the pressure losses that do not add to the mixing energy, so again, we have an engineering decision to make between these two competing variables.

I will try to dig up some data on this topic for the next post and hopefully give some data that we can ground out designs on.  Also, I will try to work through an example sp that the reader can see how easy this data is to use for you design calculations.  I also will try to find some data for the plunged holes.  Currently, I am using an equation that is given in Lefebvre's book for the plunged holes and that is all I know that is out there, but I will look for more.

Also, since I am working 12 hr shifts, every other day, due to the local impact of the COVID-19 pandemic, the posts may not be as quick as in the past.

- Chris

[finiteparts]

John, 

I just saw your post.  I will be more than happy to work the numbers on your combustor. I think it will be very instructive to the readers to work through a working design, but I would like to wait till I have finish posting on the basic steps.  That way the readers can follow through also.  I might plug the numbers into a program that I am writing for combustor design and send you the output in a PM...if I can find some time.

Also, that Lefebvre paper is not an easy one to read.  They have all kinds of units issues and lacking information throughout.  It is definitely not one of his better papers.

A few questions:  Is the pipe going to the swirlers providing the air to that region?  If so, any idea of the flow splits?   Do you know what size the leakage area at the slip joint might have been?  The swirlers were 45 degree vanes?

Thanks,

Chris

[racket]

Hi Chris

No worries :-)

Yep , the Paper wasn't the easiest for me to understand , but beggars can't be choosers, it was the only one available to me 25 years ago , I had to make the best of it, it was "way over my head" , not for this DIY'er .

The pipe to the swirler end was a later addition to increase the thrust level that was already at it maximum , didn't know at the time that a huge truck turbo could only produce 110 lbs of thrust , spent wasted years searching for the impossible, trying all sorts of different things, scrolls, wheel clipping etc etc , it was only after I "got serious" and started doing some maths that the truth came home to me that I never was going to get any more thrust , we live and learn .

Very little leakage at the slipjoint , the flametube was fixed at the fuel nozzle end and the expansion witness marks on the slipjoint snout entering the turb scroll indicated things were sliding fairly tightly with ~8 mm of differential axial expansion between FT and outer can .

The swirl vanes were from an oil burning boiler and had to be squashed "flatter" to reduce their flow area back to ~10% of total wall hole area, so probably at ~30 degrees ( 60 from axial) which would probably have produced more "spin"

Gap between FT and can was ~13-14 mm at the top half and ~25-30 mm at the bottom , airspeed at the transition area reduction would have been ~100 ft/sec heading upwards past the Secondary holes .

I really can't remember any of my thinking about the design and only have one very scrappy rough drawing ...............LOL, I didn't think I'd be still doing turbines a quarter century latter , the drawing for the bikes oil tank is of better quality .

I'll look forward to reading your Posts , they'll probably answer my queries

Cheers
John

[britishrocket]

Hi Chris,

It is good to see you posting on this topic again. I have been reading your notes along with reading Lefebvre. Regarding the annulus flow entering the holes, would it be advantageous to add a small semi circular "scoop" over each hole, facing the direction of the air flow to guide it into the hole?

Carl.

[finiteparts]

Carl,

Glad you are finding the thread useful in addition to Lefebvre's book.  That really is a must read for anyone interested in GT combustors.

Yes, I would think that the scoops would capture the dynamic pressure and "help" the orifice flow more full, but I haven't checked this out and verified my gut feeling.  These scoops and flow/swirl brakes have been used many times in the past on older designs.  I feel that this may be due to the ability to change an liner orifice's flow characteristics relatively easily with these devices...sort of like a "paste-on" fix for other problems.

Here is a resource on Cd's that covers the scoops among other things.

NACA TN 3924, "Discharge Coefficients for Combustor-liner Air-entry Holes II : Flush Rectangular Holes, Step Louvers, and Scoops"

ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19930084900.pdf

They do a very nice job of covering different scoop designs and several hole sizes, which may be a bit larger for our use.



This plot (below) does not suggest that the scoops are that great if you look at the curve based on the hole through-flow area.  I am not sure that we care what the curve based on the scoop face area.  I have only taken a brief look at this, so I can't give a educated opinion on this as of yet. 



I will try to give it a more thorough look and report back later.

Enjoy,

Chris

[britishrocket]

Hello Chris,

First of all, thank you for taking the time out to reply to my question. I am sure you have many other things that you ought to be concentrating on during these dark times.

Thank you also for pointing to the NACA reference for scooped hole Cd values. This is exactly the sort of reading I enjoy.

I have been looking at the two figures you have included in your latest post. To begin with I was not altogether sure as to why the report authors had plotted two sets of data for the scoop fitted holes, one based on flush hole area and the other based on scoop face area.

I have come to a tentative conclusion, and I would be grateful for your input, when you have time. My idea is as follows.

I think that the "thumbnail" type scoops would do very little to direct the airflow into the hole, and the resultant flow would look much like that shown in the top figure of your post from 31st March. That is to say, a relatively low Cd value caused by the vena contracta that is formed as the flow enters the hole. This is why I think that scoop face area matters. If the scoop is higher and wider, then the airflow through the anullus will be effectively turned (albeit with inevitable turbulence at the scoop inside corner) and will be more likely to enter the hole parallel to the axis of the hole. Much like the second figure in your 31st March post, showing the dimpled hole leading to improved Cd values.

I suspect that the problem would be designing a sufficiently well proportioned scoop with good flow characteristics that did not interfere with the air flow inside the anullus, and whose height did not result in an overly large anullus gap.

[CH3NO2]

Hi Chris,

Have you published in AIAA, ASME or similar? If yes post links. It would be great to see.

Tony

[finiteparts]

Apr 4, 2020 8:32:32 GMT -5 CH3NO2 said:

Hi Chris,

Have you published in AIAA, ASME or similar? If yes post links. It would be great to see.

Tony

HiTony,

No, I have not had the opportunity to publish any of my work.  The funny thing about working advanced or customer funded programs is that they tend to not want to share the leanings for some reason.  (ha!!)  

I did have the opportunity this year to act as a peer reviewer on some heat transfer papers for the upcoming 2020 ASME Turbo Expo, but do to the current enviroment, I can't see the conference actually taking place...since the idea of conferences is to bring large numbers of people from around the world together to talk shop and make connection in the technical field.  Not very wise in this social distanced climate of today.

- Chris

[finiteparts]

Apr 3, 2020 6:39:54 GMT -5 britishrocket said:

Hello Chris,

First of all, thank you for taking the time out to reply to my question. I am sure you have many other things that you ought to be concentrating on during these dark times.

Thank you also for pointing to the NACA reference for scooped hole Cd values. This is exactly the sort of reading I enjoy.

I have been looking at the two figures you have included in your latest post. To begin with I was not altogether sure as to why the report authors had plotted two sets of data for the scoop fitted holes, one based on flush hole area and the other based on scoop face area.

I have come to a tentative conclusion, and I would be grateful for your input, when you have time. My idea is as follows.

I think that the "thumbnail" type scoops would do very little to direct the airflow into the hole, and the resultant flow would look much like that shown in the top figure of your post from 31st March. That is to say, a relatively low Cd value caused by the vena contracta that is formed as the flow enters the hole. This is why I think that scoop face area matters. If the scoop is higher and wider, then the airflow through the anullus will be effectively turned (albeit with inevitable turbulence at the scoop inside corner) and will be more likely to enter the hole parallel to the axis of the hole. Much like the second figure in your 31st March post, showing the dimpled hole leading to improved Cd values.

I suspect that the problem would be designing a sufficiently well proportioned scoop with good flow characteristics that did not interfere with the air flow inside the anullus, and whose height did not result in an overly large anullus gap.

Carl, 

I finally had a few "free" moments to go through this document and found it quite interesting.  It is quite in depth and required reading through the majority to actually capture all the pertinent items that they were trying to relay.  I had not seen the pressure or boundary layer correction terms before reading through this and learning about them as applied to these Cd calcs was worth the time investment.

Ok, so I think that I will start off with Figure 19, which shows the corrected Cds based on the hole size.  (Note:  They use C, Cp, Cp, delta* to represent Cds...I will just use Cd to represent them here)



Let's start with the flow parameter that is used as the x-axis variable.  What the numerator is quantifying (Pd - pj) is the pressure drop across the liner or you can also view it as the jet dynamic pressure.  Pd is the total pressure in the duct, which we will take as the total compressor discharge pressure ( I use Pt31 based on SAE ARP 755A standard turbojet station naming conventions) and pj is the jet static pressure.  Since we know that for subsonic flow, free jet theory tells us that the static pressure of the jet must match the ambient static pressure, we can take the pj to equal the static pressure in the combustor liner,  which I designate Ps40 (plane 4.0 is the combustor exit).  Now the denominator is the dynamic pressure in the annulus (or duct as termed in the paper), (Ptd -Psd).  You can also calculate it based on the flow speed in the annulus as, q = (1/2)*rho*V^2.  Ok, so hopefully the calculation of the flow parameter is clear to all the readers.

If you are reading Lefebvre's book and get a bit hung up on his formulation of the liner pressure loss parameter, K, here is how they are related.

Lefebvre's is shown as K = (1-delta_pL/q_an).   What wasn't initially clear to me was that the delta pL term is the static pressure loss across the liner, thus...

Note:  I will use the _an subscript to mean in the annulus and the _L subscript to mean in the liner)

So from the NASA paper,  (Pt_an - Ps_L) / (Pt_an - Ps_an)  

Since Pt_an - Ps_an is just the dynamic pressure (assuming M < 0.3), q_an,        = (Pt_an - Ps_L) / q_an

Then rearranging the above definition of dynamic pressure in the annulus to find that, Pt_an = Ps_an + q_an we then get.

(q_an + Ps_an - Ps_L) / q_an = (1 + (Ps_an - Ps_L)/q_an) = (1 + (delta_Ps / q_an) which is Lefebvre's equation.

Ok...sorry if that is a bit long, but I thought others might get the initial confusion that I had over this point.

So back to Carl's question.  If you read the report, you will see that they are reporting the two different values of Cd based on the two different reference areas in order to allow easier comparison to the slot type configurations. They also mention that generally you would use the smaller of the two areas to define the reference area to calculate the ideal jet flow, but they purposely reported the Cds based on the smaller and larger areas for comparison.

I recommend using  Figure 19 and doing your calculations based on the flush hole diameter.  If  you look at figure 20, you can see that the Cd's for the thumbnails are lower as long as the flow parameters are high, but once you drop below a flow level of around 0.40 to 0.6, the thumbnail becomes much more effective.  The flow number could be low due to a local high velocity that would cause the local annulus dynamic pressure to be larger than expected.  I agree with all your statements, except is was apparent from the report that just adding a bigger scoop did not always help and of course as said before, if you have a very high local velocity in the annulus, the scoop would not produce a better jet.

Another point that can be surmised from this paper and the other papers done by NASA on these Cd test is that the local annulus velocity really begins to hurt the ability of the flow to turn once it goes above the 150 ft/s level.  In an ideal world, the annulus would be a plenum and each jet would feed evenly and produce jets that penetrate into the liner at a jet angle of 90 degrees. But our world is far from ideal.  We need flow in the annulus to cool the backside of the liner and he annulus height required to have very low velocities would be too big and heavy.  So we make a compromise.  Mellor's book on gas turbine combustion gives a good summary of designing the annulus sizing based on flow speeds and similarly inside the liner.  I happen to like this approach better due to the fact that is way more intuitive.  But either approach gives roughly the same numbers.  So if you do decide that you would rather design the flow passages based on the local bulk flow velocities, as I mentioned above, the 150 ft/s velocity that was covered in this NASA report seems like a good upper limit to the annulus flow speed.

I hope that helps,

Chris

[finiteparts]

Oh, I just realized that I forgot to cover a few things on the Figure 19 that I thought were very important.

The first thing is when you think about using scoops or thumbnails, the reason that you should be using them is because you have higher dynamic pressures in the annulus.  If you look at the plot in the previous post (Fig 19), you will see that the flush holes go to a zero Cd when the flow parameter goes to one.  This means that if the flow speed in the annulus gets high enough that the static pressure in the annulus becomes equal to the static pressure in the liner, there is no driving force to turn the flow into the liner.  Even though the equations are based on the total pressure in the annulus, you shouldn't get confused...the local difference in the static pressure is what turns the flow into the flush hole.  As the flow approaches the hole, the low pressure in the liner influences the flow in the annulus to turn in.  If the flow is going so fast as to suppress the local annulus static pressure to a level similar to the liner static pressure, then the liner can not influence the annulus streamlines at all.

To this same point, the scoops act to locally recover some of the annulus flows dynamic pressure.  This means that locally the scoops have higher static pressures that drive more flow than could have been done otherwise, and thus these design features only make sense if you have too much dynamic head in your annulus.  If you make your annulus too small and have trouble feeding the upstream holes, adding scoops might be a means of mitigation. 

As the flow travels through the annulus, each row of holes pulls off a portion of the mass flow and thus the flow speed in the annulus is reduced according to mdot = rho* V * A.  It is unusual for downstream holes to suffer from lower flow parameters and thus they are usually flush or plunged holes.  It is the upstream holes that will have the larger portion of the system airflow passing over them and thus, they are usually the ones that need the medicine to fix poor flow penetration or mixing.

It should also be noticed that if you do have sufficient annulus height such that you maintain higher flow parameters, these scoops will actually hurt your Cds.  Thus, they are not a 'fix-all" that you should apply indiscriminately, their use should be mandated by local poor flows or design constraints.

Enjoy!

Chris