Anonymous HG Poll

[americanjebus]

Pimp,... ^ i'll read that later.

americanjebus = super intestinal fortitude

Looks like I'm heading toward a 57 trim later this summer with the same set up, except for about 80 ftlbs on ARP bolts and no way I'm going over 11-12 psi. :icon_razz

Ya i am not proud of the logic behind my work because i did it all 2 weeks after buying the car and like 2 days after finding this website so i was not very informed about the ways of the 7m. But its been holding for 14k miles w/ the stock ct at ~11psi and the 57trim for a good 2k miles ~13psi. I do put the block tester on it every other week to make sure i detect any sort of leak early.

BUT i am VERY meticulous about this car from the start so i spent a week doing this HG job. I went over the block w/ a putty scraper and brake cleaner for a full day and it was just as smooth and reflective as the head after i brought it back from the machine shop lapped and cleaned. the stock head bolt thing still bothers me but they are only to 76ft lbs and i know they can hold slightly more.

I dont let the car go into positive manifold pressure unless i know EVERYTHING is working as it should. Fuel , Spark, Timing, and engine temp and good octane.

Im getting away w/ the 57trim due to the wideband and maftpro and i have been very carefull about pulling fuel and monitor my boost w/ an eagle eye. When i know my Tune is safe and ideal i have no worries about hitting the 17-18 psi. Mods are linked in sig.

and if it blows, its just an opportunity to tear it down and do it right.

 

[Imagination6788]

I answered with what I think Toyota did. Mine blew around 30K miles and under extended warrenty. I really think they need to do a recall like they did on the 4Runners.

 

[rumptis]

We are actually seeing a trend here.

 

[starscream5000]

Good posting pimptrizkit .

 

[Shytheed Dumas]

The Poll Data Analyzed

I took the “data” from the poll above and tried to work with it with little success, and then enlisted the help of a Six Sigma Blackbelt. From a statistics standpoint, it is completely unbalanced, which is to say that there are comparisons that don’t exist that should in order to say that there are differences of any significance or meaningful interactions. On top of that the number of samples per condition vary greatly, and there’s no replication to prove whether the data that are available are representative of the population or just an abnormal 'blip' in time. That’s the bad news.

So understanding that things may not be as they seem, there are a few useful things we can learn from the data by crunching it as “almost a Designed Experiment”, and by using multi-variate analysis. I’m pulling the observations from graphs generated with Minitab, but they can really get confusing and I can’t get to a website that can host them from work anyway, so I’m not including them.

So without further ado, here are a key few points:

1. Not machining your head will DOUBLE your chances of “problems”, regardless of HG type or what you do with the block, and overall you are better off using a metal HG.
2. Ignoring whether you machine the head and block or not, composite HG’s fail about 22% of the time compared to about 18% for metal. The significance here is questionable.
3. The surprising thing is that there is no significant difference between machining the block and not (disregarding type of HG or what you might do with the head). Argue this if you like, but that’s what the data tell us.
4. If you leave your block unmachined and have your head machined, it really doesn’t matter much if you use a metal or composite HG (somewhere around 11 +/- 2% chance of failure overall).
5. If the head is left unmachined with an unmachined block your chances of failure increase from roughly 20% to roughly 37% if you use a metal HG, ergo a word of wisdom: If you aren’t going to machine your head or your block, DON’T use a metal HG!
6. Here’s a weird one: If your block AND head are both machined, you TRIPLE your chances of failure by using a composite HG! (10% failure for metal and ~34% for composite). Compare that to observation #2: Block unmachined, head machined = no difference in HG and #4: Block machined, head machined = USE METAL HG.
7. What’s lacking is what would happen if the block was machined and the head left untouched. It’s unlikely anybody would do this, but for the sake of understanding interactions it is necessary to find out.

So that’s the best we could do with a less than ideal data set. I wouldn’t take all of this to the bank, but there are probably some good indications of what happens when we make decisions on how to do our HG jobs. The only way to improve what we learn is to improve the quality of data, and this would be best done with a DOE (design of experiment) with at least two replicates and numerous repeats with a better response than “problems”. Not likely to happen, so hopefully picking apart what data we do have is at least a little helpful.

-Kurt

 

[dugums]

Nice post.

I was hoping at least to have more data and perform similar evaluations. Obviously, the quality of the data is a little suspect (based on the vagueness of "problems" etc.)

As I was creating the options for the poll - I originally had a lot more possible responses. When I looked at it, it just seemed like too much - that I would never get enough people to read through a whole page of choices - hence why I distilled it into what it is.

The other problem is, I know more people who have blown head gaskets on here then have responded. Quite frankly, the next ten people to respond might report failure and completely change the way this data set appears.

I think we all are in agreement that properly machining the block and head are the best way to do it. Most of the people who have had issues with that setup had bad machine work, forgot to machine the timing plate, or were running more boost than the average enthusiast might.

The main question I was searching for an answer for myself was if it was worth the considerable risk of doing a MHG on a block that hadn't been machined. Well, it seems like it just might be based on the statistics.

Anyway - keep the responses coming - the more responses, the better the data... And please post any experiences with failures - I think that is what we are all wanting to hear.

 

[Shytheed Dumas]

I think we all are in agreement that properly machining the block and head are the best way to do it. Most of the people who have had issues with that setup had bad machine work, forgot to machine the timing plate, or were running more boost than the average enthusiast might.

The main question I was searching for an answer for myself was if it was worth the considerable risk of doing a MHG on a block that hadn't been machined. Well, it seems like it just might be based on the statistics.

I would agree. The BEST situation was metal HG with both block and head surfaced, but if you aren't going to machine the block the results here say you can go ahead with a metal HG.

What's really interesting is that virtually everybody has said that it is almost gauranteed that you will blow a metal HG if you don't machine the block, which is really a value statement. The data disagree... so far. Based on this I wouldn't necessarily discount using a metal HG, but you can be sure I would spend a good deal of time prepping the hell out of the block myself.

Any 'gotta machine the block to use metal HG' diehards want to comment?

 

[upgradedsupra]

What about the option of lapping after machining?

Duane

 

[Shytheed Dumas]

Yeah, soooo many ways to look at this.

So here's the way a proper design would look, omitting Duane's suggestion because it creates an unbalanced study and requires more advanced stats with more runs

* 2^3 Designed Experiment = 8 runs, probably need 3 repeats per run, and at least one replicate of the whole thing = 48 HG jobs total, and layed out like this:

Variables:
HG: Composite (-) and Metal (+)
Head: Unmachined (-) and Machined (+)
Block: Unmachined (-) and Machined (+)

Runs Expressed in terms of HG/Head/Block
1. +++
2. ++-
3. +-+
4. +--
5. -++
6. -+-
7. --+
8. ---

So there are the 8 runs with 3 engines a piece, and when you're done, repeat the entire thing at least one more time. All would need to be run with identical mods, boost, etc, and evaluated for leaks and failure over time.

Complete and balanced designed experiment with stastical confidence in what really matters and how these things interact to predict success. Anybody game?

 

[dugums]

Complete and balanced designed experiment with stastical confidence. Anybody game?

Ha.

I am doing mine right now and have been very tempted to try a MHG despite the fact I have only machined the head. The car doesn't seem to have ever overheated (I was doing the HG as maintenance) and the shop who did the head said it was within factory spec.

I had them prep it for a MHG, though it is not a "mirror" finish. The guy who did the work was a strong proponent of Cometic gaskets and when I told him I was probably going to end up just doing a composite, he said I might want to think about trying a MHG even if I wasn't going to machine the block.

I don't think I have big enough balls to go through with it though:nono:

 

[dugums]

Variables:
HG: Composite (-) and Metal (+)
Head: Unmachined (-) and Machined (+)
Block: Unmachined (-) and Machined (+)

Runs Expressed in terms of HG/Head/Block
1. +++
2. ++-
3. +-+
4. +--
5. -++
6. -+-
7. --+
8. ---

Who the hell is going to do #3 and #7, lol.

 

[Shytheed Dumas]

Who the hell is going to do #3 and #7, lol.

lmao. You've gotta go through the whole thing to understand it.

I say go for the MHG with the machined head and prepping the block yourself. Even though this isn't the best data set in the world, if I had it to do again I think that's what I would be doing now. (disclaimer: if you try this and it fails it's your ass! :biglaugh

 

[dugums]

(disclaimer: if you try this and it fails it's your ass! :biglaugh

Like I said, I don't think I have the cajones. I have gotten so much help with this one - I would hate to try to do it again.

You can see how fast the data can change though - the vast majority of recent respondents have been answering "block+head machined - no problems".

To evaluate my decision - I was hoping for more answers on the composite side.

lmao. You've gotta go through the whole thing to understand it.

Yeah, I know we would need samples for every combination of factors - We would need to convince some unsuspecting young lad that prepping the block and not the head is the way to go if you're doing a MHG... Oooo - and another one to try the composite, lol

 

[soapra]

Metal HG, Head + Block Machined -- No Problems


Only way it should be done!