Exhaust Header Size for My Rebuilt 2.7L Motor
 

I decided maybe it would be best to start a new topic thread to continue the discussion about the best header OD size for my recently rebuilt 2.7L motor.

We (mb911, mikedsilva, scarceller, winders and others) have been having a very good, enriching, knowledgeable discussion on this topic over on my "CIS to ITB EFI conversion" thread.

But that thread has run it's course as a "how to" for like minded folks that want to successfully install Al Kosmal's X-Faktory CIS to ITB EFI conversion kit.

As a recap, I decided towards the end of the rebuild of my 1975 911 Carrera 2.7L engine to employ the "bigger is better" philosophy in the selection and installation of 1.750" OD exhaust headers. All of the more experienced members of this forum quickly pointed out the 1.750" OD headers were way too big for my little 2.7L motor and that 1.50" OD headers were most appropriate.

Well, I had the 1.750" OD headers (Bursch Exhaust BX.9258.SS) installed so I ran with them for the initial start up, 1000 mile break-in. and dyno tuning run.

The dyno tuning results clearly showed there was a problem: a "hole" in the torque curve between 3000-4000 rpm.

http://forums.pelicanparts.com/uploa...1723254202.jpg

That dyno result was not acceptable to me, so I have ordered a set of 1.625" OD headers (Bursch Exhaust BX.9158.SS) and a set of 1.50" OD headers (Dansk 1620800470 & 1620800480) which I will install and run on the dyno once I receive them. Then based on the dyno data, I will decide which size header (1.50" OD or 1.625" OD - I have already conceded that 1.750" OD is not right) to run given my intended use for the car as a weekend canyon carver.

I will receive the 1.625" OD headers tomorrow, so they will be the next exhaust option that I will run on the dyno (my next dyno appointment is Aug 28th), and I will share the dyno data as it becomes available.

Dan,

I am running some 1 3/4" custom headers William Knight has made so I don't need another set of headers. Plus, I am not a fan of the Bursch headers in any O.D.

Here is Williams contact info:

William Knight
knightrace@mac.com
615.969.4917

Dan the 1.625 headers will work better for sure. I want to put this information down for others to see so they can reference it.

ALL tubing is measured OD

ALL pipe is measured ID until over 12” ( that is of no concern to us folks)

ALL headers when advertised size are referring to OD. In my many years of doing this and hundreds of sets made/sold I have never seen it different. For example the original SSI was 1.5” with an ID of 1.375 because a wall thickness of .0625 and no they weren’t metric tubing and most the headers you see are not metric tubing.


I hope this helps someone else

Hi Ben,

Thank you, that is good info for all of us.

I will add that when I was looking on the internet for off the shelf 1.50" OD headers without heat exchangers yesterday, the vast majority of the units I saw advertised were 1.625" (41mm) and advertised in US measurements.

The only exception to the tube size in OD rule I saw was from Patrick Motorsports (Porsche 911 Exhaust Header Set With Street Adapters - 1-1/2 Inch (1.50") (EXH P911 112SA)) and those units specifically state "Exhaust Headers - Porsche 911 - 1.50 in. / 38mm I.D. Primaries - Mild Steel - includes Street Muffler Adapter Set" in their description.

Dan I am 99% sure that PMS is identifying OD then ID in that description which makes it even more confusing.

I also am almost certain William Knight is having his headers made elsewhere but that could be a bad rumor but regardless those are quality units.

Yes , if PMS unit is 1.50” OD , then their description is very confusing as 1.50” = 38mm.

It is for sure. Have you looked into MSDS headers and having them ceramic coated. If you call Marty he is extremely knowledgeable and has been doing this since the 70s

We've been designing and building custom headers specific to application for years.
There is a lot more to achieving an optimal torque curve and HP number than OD tubing size. Muffler design is also extremely important, we build the entire exhaust as one system where the flow is optimized from exhaust port to tailpipe. You can't expect to slap on a set of generic headers and a one size fits all muffler and have anything good come of it.
Here is the info we request:

Displacement
C/R
Induction
Cam profile
Head work and I/E port sizes
Redline RPM
Expected torque curve and peak RPM
Expected HP and RPM
How the car will be driven and if it is track or street or both
How loud and where do you want the tailpipe(s) to exit
Do you need heat exchangers

With this information a properly optimized system can be designed and produced. I cannot imagine spending the money these engines cost to build today and leaving so much on the table using an off the shelf exhaust system.

.

You want to maintain a balance of capacity between the intake event, the combustion event and the exhaust event.

Here's a comparison of flow volume at 100% VE for the different sized combustion events, notice that a 2.7 @ 8k flows the same as a 3.8 @5.6k
http://forums.pelicanparts.com/uploa...1723323256.gif

a close up of the above
http://forums.pelicanparts.com/uploa...1723323431.gif

when sizing the intake or exhaust the idea is to keep flow up but also to though put the necessary volumes, too big a flow velocity drops too small and unwanted pressure builds

the exhaust tube length and width is also a tuning tool to constructively put a neg. pressure wave at the exhaust port @ TD compression overlap when both valves are off their seats, The more overlap the more constructive this signal is. The purpose here is to draw extra gasses out of the exhaust port and to draw extra gasses in through the intake port. But this is only efficiently done w/ open headers w/ the right geometry and topology. This neg. acoustic signal strength is solely determined by the cross section area of transitions in the exhaust, the largest being at the collector outlet in an unmuffled system.

http://forums.pelicanparts.com/uploa...1723323990.jpg

muffled systems have this refracted wave but in weakened form and from smaller cross section transitions found in the exhaust.

There is also a destructive rarefication presented to the exhaust @~2k rpm, this one inhibits flow, it can usually be detected by a drone in the exhaust noise.

different exhaust cross section areas mostly move the torque peak up or down a few hundred rpm a long as they are close to what's need to maintain flow velocity, the gasses leave the exhaust port @`~120m/s, you want to keep that velocity up despite the slowdown caused by cooling and expansion
http://forums.pelicanparts.com/uploa...1723324479.gif



length of the primary tubes merely rocks the the torque curve around the torque peak
http://forums.pelicanparts.com/uploa...1723324874.gif

to actually move the torque curve and change it's shape you need to play w/ cylinder pairings, using what's called Tri y merges vs the usual 3 or 4 into 1 collector, it's easier to see w/ a 4 cyl or a V8 than w/ a flat 6 because the flat 6 has ideal pairings w/ the 3 into 1 we mostly all use.
http://forums.pelicanparts.com/uploa...1723325156.gif

to get a similar result from a flat 6 you need to do something like this

http://forums.pelicanparts.com/uploa...1723325244.jpg

Dan is that centerline measurements? Pretty long headers

Centerline is the best way. You can take a piece of string and hold it right where it goes into the collector, then work your way down the tube and once at the flange mark the string and lay it out flat on a table and measure.

Critical aspect of measuring tube length at the center line is to replicate the pathway of the center of the tube by staying in the center of the bend radius, not going to the outside or inside of the bend. This may seem obvious but may not be to everyone. The spec we use on primary length is +/- 1/4" to remain true equal length.

VE Table After 1st Auto Tune Session With 1625 Headers
 

This morning I took the car out for an auto tuning session after installing the 1.625" OD headers. I ran the car mainly on the freeway at idle to about 4800rpm and mainly 0 to 60% TPS with a couple of WOT pulls.

Difference Report from TunerStudio of the After (left VE table) versus Before (right VE table).

http://forums.pelicanparts.com/uploa...1723401363.jpg

On the right side table (1.750" headers) you can see most of the VE cell values increased (red numbers) with the 1.625" headers in the target region.

Since the AFR targets have not changed, increasing VE cell values means air flow at the rpm / load for those increased cells is also increasing. That means when I run the engine on the dyno with the 1.625" headers, the torque should also increase (more air & fuel, bigger bang, more pressure).

There will probably still be a dip in the torque curve at 3000-4000 rpm, but it should be less. Then I will install and run the 1.500" OD headers and see what that data shows me.

This is great VE comparison data on the two headers - thank you for posting the analysis.

It looks like you have a nice bump 3k-4.5k with the new headers --- and even better, no loss on the top end. I look forward to seeing the dyno and 1.5" header comparison.

Thx Ben for the MSDS,inc. honorable mention !
We mfg this 'Merge collector style>> much like the way the factory built their competition racing headers back in the '70s. 1.625" od is correct for a 2.7L; and as a side note we mfg the three popular sizes 1.500" 1.625" and 1.750" od's...and No upcharges for the larger sizes either.
Three styles of 1,650°f Ceramic coating availble.
www.msdsinc.comhttp://forums.pelicanparts.com/uploa...1723437197.jpg

That is not what Ben said to do. the method you are using is not close to accurate. Look here:

https://forums.pelicanparts.com/911-engine-rebuilding-forum/940145-how-primary-length-measured.html#post9406586

The key is to keep the string at 90% of any turn...this means you may not be able to measure it without measuring in sections.

Yes, thanks for posting this! The information posted by the OP (so far!) has created the single most valuable header diameter thread that I've seen.

I also was very interested to see the lack of change in the upper range of the ve tuning, but it was unclear if the autotune session included any activity in the upper RPM ranges?

Hi Jonny,

When I ran the auto tune session I stayed in the 2500-4800 rpm range to focus on changes in the 3000 - 4000 rpm region of the VE table - so cell values in VE table above 5000 rpm haven't seen the effect of the 1.625" headers.

The string method is close enough. The difference of an inch or so will make very little difference in the computation of resonances in the exhaust primaries.

Hopefully, 3rd time gets me in the ballpark. This time I measured each welded section individually from center of weld to center of weld, positioning the header so the I was looking down on the section, orthogonal to the plane of the curve of the section, and used the string to trace a path along the centerline of the section. Then I added the sections together to get the length of the primary from the head flange to the end of primary in the collector.

1.750" OD header - #1: 34 3/8", #2: 33 1/2", #3: 33 1/8"

So I measured the primary tube lengths for 4/5/6 bank header using the 90 degree centerline method described in my last post.

To recap: Measurements were taken on the Bursch BX.9258.SS 1.750" OD headers.

Primary tube length was measured from the head flange to the end of the primary tube in the collector. I double checked / remeasured several tubes to check the repeatability of my measurement technique which was within 1/4".

#1: 34 3/8"
#2: 33 1/2"
#3: 33 1/8"

#4: 35 3/8"
#5: 36 1/2"
#6: 31 1/2"

I am not a fan of the Bursch headers....I would talk to William Knight about headers....he knows what works and doesn't work for your build...

Dan
I really like the way you are approaching this evaluation. All factual.

A few years ago, I went through a 'similar' experiment and learned the value of the exhaust gas flow volume vs speed.
My 3.0 itb engine ran a set of 1&5/8 euro headers and I wanted to run a GT3 muffler. So I manufactured the collector pipe from the headers to each side of the muffler.
First set was 2 inch pipe. On first drive, the car simply took off like a rocket! It really felt like I had a turbo down low... but at 5000rpm it seemed to run out of breath.

Second set was 2.5 inch... the sound was nuts. And the torque down low had all disappeared. In fact, it felt like an absolute slug. Top end revved out well to 7k+.

Third set was 2.25 - this gave me most of the bottom end torque back and did not restrict the top end at all.

After I did this, William explained to me why it happened and yes, I could have avoided the hassle by just calling him first, but the first hand experience is worth it.

You are way overthinking this
@~225hp and <7K rpm you are going to be fine w/ SSI type headers 1.5" OD(1 3/8" ID) if this is a race car where you are aways going to be in the 5-7K range then maybe 1.625"(1.5" ID)


over size hurts more at the usual sub 5K range than undersize hurts 6-7k

I know people that run 1.5" OD on street 3.6 because under 5k they enhance performance, they almost never go to 7K and could care less up there.

The primary pipes have a fairly wide range of acceptable differences in length w/o adversely affecting power.

1-5/8" secondary pipes seems small, is this a typo?

Very interesting to me that this made such a noticeable difference. I guess there's a reason why the secondary pipes on street setups are almost always 2". When you think about it it they only need to deliver the flow from one primary pipe at a time.

1000% agree and that was my point on the other thread as well. Dan do the KISS method on this. Buying different cams is fine if you are craving more power but your kind of chasing your tail

Why on earth would you go to the trouble and expense of installing a 3.6L engine in an earlier chassis just to neuter it to 3.0L output levels? A properly designed system will preserve the torque curve AND release the horse power.

Many of the systems we design and build are for racing applications where they are looking for that last 5HP. Even with a stock engine you can't get to the engine's full potential using off the shelf systems. Add modifications and the diminishing returns stack up even more if the flow is not optimized to compensate. Decision becomes how much compromise is acceptable.

Just going to throw this out there, I have run 1.75" headers on my basically stock 2.7 CIS for 5k miles, mainly because I owned them already and I like the sound.

On one hand, I know it's slower than if the primaries were the proper size. On the other hand, this is a street car, not a race car, so if it's a .5 second slower, it honestly does not matter, especially if it sounds glorious at higher RPMs

If you're tracking and driving competitively, all the info above makes sense, but idk if I'm getting old or what, but I just don't care about performance anymore, especially with all the police around that hate anybody having fun. I'm all about experience > metrics now.

All you are doing is hurting your driving experience. 1.5" headers, which is what you should be running, would not sound any different assuming you used the same muffler.

But you go right ahead and do what you think is best....

I agree and I wouldn't do it either other than as a temporary affair, Which I did w/ my first 964 3.6 t/p, around town you never notice the difference, it's only strangled and detrimental up top.

💯 agree with that statement

If someone wants to straight up trade me for a similar system (headers with heat boxes) I'm interested, but until then I don't care.

you're right! I meant to say 2.25.... I'll go up and edit my post.
thankyou!

A collector for a 911 6 doesn't have to be huge as the pulses are well separated
@7500rpm there are ~1.5 to 1.8 pulses in a 38" primary at any given instant, how many depends on the rates of cooling and expansion.

each collector gets 188 pulses /sec
each pulses travels from ~.5 to ~.6m/s
this means that a given pulse clears the collector before the next arrives

there is on average ~.3 to ~.4 pulses in a given collector at a given instant.

That’s cool….but I’d care a lot if the bottom end of my engine was as compromised as yours is…you are tossing away a bunch of low end.

each collector gets 188 pulses /sec
each pulses travels from ~.5 to ~.6m before the next pulse enters the collector
this means that a given pulse clears the collector before the next arrives

I got distracted and didn't correctly type the line underlined

In that part I was solely refereeing to the collector which for the purposes of the above calculation was 10" long

gas velocity was assumed to be 120m/s, 100m/s is probably closer due to the decel of the gasses as they pass through the system, 120m/s when they leave the exhaust port.

As a PS to the above
I never measured any of the collector lengths for the headers I had in hand

all had 26 to 29" long primaries including the collector from 1.5 to 1.75" OD these are all quick and dirty ests.

the outlier was 993RSR header @19" long

bump

I am sorry Dan was there a question? It seems Bill, Brian, Winders shared much valuable information and your car and VE tables support the smaller headers helped your car

No questions.

I'll have dyno data to share in 9 days that should support the conclusion that my 2.8L engine performs better with 1.625" headers than 1.750" headers.

But after nearly 250 miles in the car since switching to smaller headers the one thing that disappoints me most with my 2.8L engine build is that the engine has lost ALL power between idle and 2000 rpm, so driving in town or stop and go traffic on the highway is a royal pain in the butt - revving the engine, slipping the clutch repeatedly to get the car rolling is unacceptable and smaller headers have done NOTHING to improve that situation. Once the car is rolling and out on the open road (or track), it goes great.

The real problem I am having isn't the header size selection, it's the camshaft selection I made - the WebCams 911S MFI cam (p/n 05-062) is not good, according to William Knight it's not enough exhaust flow to balance the intake flow. William guarantees that I will make more power across the rev range and the car will be easier to drive with the K45 cams.

So my next move is replacing the 911S MFI cams with William's K45 cams and a set of his headers. That dyno data comparison will be posted elsewhere when I have it.