CR engine HPFP analysis

[torqueit]

And if you look at a Ford enthusiast site, you see the exact same discussion about their HPFP's.

Multiple revisions, early failures, etc. ....

 

[Ski in NC]

It should be self stabilizing. Imagine for some reason the roller is cocked before cam starts to lift. The first part of the roller to contact the cam will tend to rotate it back into alignment. I don't think an anti-rotation feature is needed, as thousands of functioning hpfp's prove. I think any anti-rotation feature would come into play once some other sort of failure began.

However, if I designed this thing it WOULD have an anti-rotation feature.

I don't like the roller without an axle. An axle with smaller diameter would make it much less likely for the roller to skid on the cam. If the coefficient of friction was the same on axle and roller od, it still would not skid. Without an axle and with identical COF, roller will skid.

Relying on a coating to reduce COF on the bucket side of the roller compared to the line contact on the cam, that is sketchy at best.

I think this may be the fundamental flaw in the design. No axle. Roller skids on cam and wear products are created. Downhill from there as the swarf migrates.

Look in a VE pump. Rollers have axles.

Look at any heavy duty injection pump. Rollers have axles.

Look at any heavy duty diesel valve operating cam foller. Rollers have axles.

 

[Niner]

It should be self stabilizing. Imagine for some reason the roller is cocked before cam starts to lift. The first part of the roller to contact the cam will tend to rotate it back into alignment. I don't think an anti-rotation feature is needed, as thousands of functioning hpfp's prove. I think any anti-rotation feature would come into play once some other sort of failure began.

However, if I designed this thing it WOULD have an anti-rotation feature.

I don't like the roller without an axle. An axle with smaller diameter would make it much less likely for the roller to skid on the cam. If the coefficient of friction was the same on axle and roller od, it still would not skid. Without an axle and with identical COF, roller will skid.

Relying on a coating to reduce COF on the bucket side of the roller compared to the line contact on the cam, that is sketchy at best.

I think this may be the fundamental flaw in the design. No axle. Roller skids on cam and wear products are created. Downhill from there as the swarf migrates.

Look in a VE pump. Rollers have axles.

Look at any heavy duty injection pump. Rollers have axles.

Look at any heavy duty diesel valve operating cam foller. Rollers have axles.

I agree with you 100%. If you want the cam follower roller to rotate, you design for it to rotate laterally. If the needs of the cam roller follower absolutely needs to stay in alignment with the cam surface, you design so that at all times, 100% of the time, it stays in alignment, and can not possibly get out of alignment. That's just common sense engineering and design. I see that as a flagrant defect in design parameters, be it by Bosch or VW, for cost cutting measures.

Someone here posted up that valve spring on a motorcycle cylinder head running at various rpms, and the ocillations of the spring at certain rpms caused the spring and the valve to rotate during operation. You can see it on the motion picture, the way the spring decompresses at various locations of the spring causes the rotation, probably due to manufacturing variances in the return rate of various sections of the spring.

 

[Ski in NC]

Good point about strange spring dynamics.

We need to keep in mind the speed that this thing runs. 1X crank speed. Up to 5000rpm. Since 4stroke cams run at half crank speed, it's like the motorcycle valve spring at 10,000rpm.

Other injection pumps run at half speed. VE runs 2500rpm max.

RPM is not alone a relevant way to evaluate cam operated machinery, though. Accelerations, velocities, loads and stresses will vary with rpm, but also with other design elements.

But that roller needs an axle!!

Heck, just convert a VE pump to a CR pump!! How bout that for fun?

 

[specsalot]

It should be self stabilizing. Imagine for some reason the roller is cocked before cam starts to lift. The first part of the roller to contact the cam will tend to rotate it back into alignment. I don't think an anti-rotation feature is needed, as thousands of functioning hpfp's prove. I think any anti-rotation feature would come into play once some other sort of failure began.
However, if I designed this thing it WOULD have an anti-rotation feature.
I don't like the roller without an axle. An axle with smaller diameter would make it much less likely for the roller to skid on the cam. If the coefficient of friction was the same on axle and roller od, it still would not skid. Without an axle and with identical COF, roller will skid.
Relying on a coating to reduce COF on the bucket side of the roller compared to the line contact on the cam, that is sketchy at best.
I think this may be the fundamental flaw in the design. No axle. Roller skids on cam and wear products are created. Downhill from there as the swarf migrates.
Look in a VE pump. Rollers have axles.
Look at any heavy duty injection pump. Rollers have axles.
Look at any heavy duty diesel valve operating cam foller. Rollers have axles.

Well said - A great summation of current thinking. I'll bet these exact sentiments were the basis of VW's discussion with Delphi.

I agree with you 100%. If you want the cam follower roller to rotate, you design for it to rotate laterally. If the needs of the cam roller follower absolutely needs to stay in alignment with the cam surface, you design so that at all times, 100% of the time, it stays in alignment, and can not possibly get out of alignment. That's just common sense engineering and design. I see that as a flagrant defect in design parameters, be it by Bosch or VW, for cost cutting measures.
Someone here posted up that valve spring on a motorcycle cylinder head running at various rpms, and the ocillations of the spring at certain rpms caused the spring and the valve to rotate during operation. You can see it on the motion picture, the way the spring decompresses at various locations of the spring causes the rotation, probably due to manufacturing variances in the return rate of various sections of the spring.

Your points make perfect sense. This design is cheaper to manufacture. It looks like an elegant solution until the pump begins to circulate wear materials.

Once the roller starts skidding rather than rolling, any loss in parallelism with the cam surface will rotate it 90 DEG in short order. At that point there are no forces present capable of bringing it back into alignment. The roller is now doing point contact.

Of course the point grows larger as the roller begins to carve a grove in the cam surface - wear material cascades through the pump. The wear in the cam surface stabilizes the new failure alignment. That is how BrokenTDIs pump lived out it's last hours.

The rotational forces generated by the spring are a good observation, but they are likely not consequential in the scheme of things. Many commercial engines utilize valve rotators to help keep sealing surfaces clear / clean. Even with such purposefully designed components valve can stop rotating. My point is that the forces generated are much less than the corrective forces that keep a rolling roller aligned with it's cam.

Per SKI, it's the change in COF due to coating loss that really gets the ball rolling. Bosch kept the roller OD small to minimize contact area. Delphi (having likely been briefed by VW) went with a larger diameter roller (which likely improves lubrication) but chose to restrain it from rotation. Clearly lessons learned at Bosch's expense with VWs help. Both pumps rely on coatings so neither one is probably immune to swarf (or chemical) induced coating failure. Because Delphi doesn't run a steel piston in an aluminum bore their design looks superior.

Inspecting and replacing the follower assembly if there is coating wear / failure in process could represent a useful strategy until the manufacturers address these issues. As would finding a way to conduct in situ cleaning of the LP cavity.

 

[Niner]

Good point about strange spring dynamics.

We need to keep in mind the speed that this thing runs. 1X crank speed. Up to 5000rpm. Since 4stroke cams run at half crank speed, it's like the motorcycle valve spring at 10,000rpm.

Other injection pumps run at half speed. VE runs 2500rpm max.

RPM is not alone a relevant way to evaluate cam operated machinery, though. Accelerations, velocities, loads and stresses will vary with rpm, but also with other design elements.

But that roller needs an axle!!

Heck, just convert a VE pump to a CR pump!! How bout that for fun?

1/2 speed for the VE, 4 rollers for the VE, 4 cam surfaces providing lift to drive the plunger at the same time, 1200 bar for the VE, not 2000 + bar. I see stress analysis was definitely lacking in this new pump, like it was designed for failure, so much of it is over taxed by that one cam and one stupid little roller.

I am beginning to think that Bosch's idea of pushing the pressure parameters to new higher limits on the fuel is the defect in design. There may very well be barriers, or limits, to the amount of pressurization designs can withstand with current technology and materials.

 

[Honeydew]

Here are some interesting pdf data sheets from the Bosch Germany website showing three different common rail systems: CRS1-16, CRS2, CRS3. The Chinese get a water separator!

 

[JKC_NC]

Here are some interesting pdf data sheets from the Bosch Germany website showing three different common rail systems: CRS1-16, CRS2, CRS3. The Chinese get a water separator!

Give me the CRS3..."service life" of 300,000-400,000 km vs the CRS1-16 & CRS2 with a 150,000 km "service life".

What system do I have in my 2010 I bought in Oct 2009? 150,000 km is just 90,000 miles and not at all acceptable for a modern vehicle.

 

[flyboy320]

What's disturbing is the last two failures from this thread have used additives (at least one right from the dealers lot), and they still have experienced a HPFP failure.

 

[Ski in NC]

Give me the CRS3..."service life" of 300,000-400,000 km vs the CRS1-16 & CRS2 with a 150,000 km "service life".

What system do I have in my 2010 I bought in Oct 2009? 150,000 km is just 90,000 miles and not at all acceptable for a modern vehicle.

Don't read too much into "service life" or "design life". Consider that a car engine at lifetime average speed of 45mph and 2000 rpm goes through 2.6 billion revolutions in 100kmiles. There is no way to engineer a machine that can do 2.6 billion cycles that also can't do 5.2 billion cycles (200kmiles). Once your design is good for a billion, it is probably good for 10 billion. Unless some event "kills" it (not wear), and that is the most likely demise of a car engine.

How many car engines have you seen that failed because it simply "wore out"? Not many. Most that failed were killed in some unfortunate event.

Terms like "design life" and "service life" usually are defined in some sort of regulatory realm and have little to do with how long the machine will actually last.

 

[oxford_guy]

The Chinese get a water separator!

lol...

 

[HeAvYfUeL]

I have been following and reading this thread on and off since it started.
Lately I have been looking at what mileage the vehicles have when the HPFP fails. (I had some issues with mine after the first fuel filter change).
What I find very obvious is that they the HPFP fails:

• On new cars with only 1000 or 2000 miles. (new filter).

• Or after the first/second/etc. fuel filter change.( also with in 1000-2000 miles after the filter change).
  
  Could the priming of the fuel filter or contamination of the system when replacing the fuel filter have anything to do with it. Maybe air in the system or fuel control valve is giving plausible signals to the ECU which causes pressure pulses in the CR that affect the HPFP function. These pressure pulses affect the LP part of the HPFP and the pump slowly disintegrates. This kind of real life scenarios are probably not simulated on the test bench.

 

[Scratcher]

Could the priming of the fuel filter or contamination of the system when replacing the fuel filter have anything to do with it.

Only if the tech was trying to prime the filter by cranking instead of using VCDS in which case the pump would operate dry for a short period before fuel reached it

 

[N2UADTDI]

I don't see how the HPFP will be dry after changing a filter. The pump itself does not drain out. The moving parts of the pump will always have fuel there. But when the fuel does start flowing there will be air that gets pushed thru.
Does damage occur at this point at such low RPM? I don't know?

Being that water will tend to be in the bottom of the fuel filter canister I think it is necessary to completely clean the canister. That will leave the entire fuel filter without fuel. This is where priming should be done with the laptop software to turn the electric pumps on.
But what do the VW techs do? From what I've heard on this forum they pull the filter cartridge out and plop the new one in. That leaves possible crud in the bottom of the canister.
What's my take on the failures? I think VW is right. It's a fuel contamination issue. If the fuel does not meet the specs the pump was designed for it will fail. How much fuel in the US does not meet these specs? Someone posted a chart of fuel scar testing from around the country. It was scary how much fuel did not meet this particular spec.

 

[sgoldste01]

Which is precisely why I add 1 quart of B100 to each tankful, bringing the tank to B2. At least I then know that my fuel meets the lubricity spec.

 

[nhdude]

Optilube XPD goes in every tankful... hopefully this helps ensure that the fuel meets lubricity spec. If B100 were available anywhere nearby, I'd use that instead. While this MIGHT help avoid an issue with my HPFP, it certainly does not guarantee it and VW should still have utilized a design that is compatible with the fuel available in the markets into which the engines are sold.

 

[sgoldste01]

Agreed. But since there's a small chance that the HPFP design might not be compatible with any tank of fuel I might receive, I use the B100 as a way of (hopefully) decreasing the odds of a problem. It's all that I can do as an individual consumer. Should I need to take this precaution? No. But I do it anyway.

 

[oxford_guy]

Which is precisely why I add 1 quart of B100 to each tankful, bringing the tank to B2. At least I then know that my fuel meets the lubricity spec.

Some here have suggested that when people add things to their fuel they don't get mixed well enough. There was a debate about it. I suppose the safest thing is to get B1-B5 at the pump—to ensure thorough mixing.

 

[sgoldste01]

Some here have suggested that when people add things to their fuel they don't get mixed well enough. There was a debate about it. I suppose the safest thing is to get B1-B5 at the pump—to ensure thorough mixing.

Well, that would be my first choice too. Here in NYS, though, we don't get any biodiesel blends from the pump. We only get D2. Luckily, I have easy access to B100, so I buy a few gallons of that every year and splash blend my own B2.

If I lived in PA or another state that mandates a biodiesel blend, then I wouldn't bother adding my own.

Having said that, I'm glad I don't live in IL, where (if memory serves me correctly) B10 is state mandated, exceeding VW's B5 max. But that discussion is in another thread, so we probably shouldn't get into it here.

 

[Westro]

Having said that, I'm glad I don't live in IL, where (if memory serves me correctly) B10 is state mandated, exceeding VW's B5 max. But that discussion is in another thread, so we probably shouldn't get into it here.

Minnesota is Mandated B10 in May 1, 2012.... I wonder how they can sell a TDI in Minnesota?

We are currently mandated B5 summer and B2 winter, and I've been in Burnsville VW when they have rolled through a HPFP failure.

 

[emdeeaitch]

What's my take on the failures? I think VW is right. It's a fuel contamination issue. If the fuel does not meet the specs the pump was designed for it will fail. How much fuel in the US does not meet these specs? Someone posted a chart of fuel scar testing from around the country. It was scary how much fuel did not meet this particular spec.

The NHTSA's line of questioning on this is dead on. If you read the Oct 7 letter to VW, it is clear that whoever at the NHTSA wrote the questions most likely thinks that VW was negligent specifically for not designing a pump that could handle the known fuel variability that occurs in the market, which they easily could have tested before coming to market instead of after coming to market. Correct me if I'm wrong, but I believe the ULSD was hitting the shelves in the US en masse back in 2006, which was 2 years before one of these cars hit the streets in US. Two years might not represent a full dev cycle, but certainly enough time to test for and detect this problem. Also, it is common knowledge (as it has come out in this forum) that fuel transport practices in the US will guarantee frequent low-percentage contamination due to, among other factors, use of tankers that carry gasoline and diesel alternately with no washing in between loads. On top of that, misfuels are going to happen due to user error, and the pump should be robust enough to handle at least one of those without massively failing--and indeed some have, while others fail massively off lower percentage contamination. The issue would be largely, though not completely, mitigated if VW had just put properly filtering in the lines to localize the pump implosions to the pump. The safety issue of instant power loss would remain, but at least the car would not have a ridiculous full fuel system replacement. Think of all the lost man hours and all the energy that goes into producing all those non-pump parts that should not have to be replaced during at least the first 200K miles of a car's life. The VW position here is totally indefensible when you consider all these factors. And they have admitted it by both switching to Delphi for the later Polo engine and by redesigning the fuel system for the new US Passat, despite the fact that it is otherwise the same motor. There is no reason for you to defend VWoA, because VWoA has already admitted guilt by their actions.

 

[oxford_guy]

And they have admitted it by both switching to Delphi for the later Polo engine and by redesigning the fuel system for the new US Passat, despite the fact that it is otherwise the same motor. There is no reason for you to defend VWoA, because VWoA has already admitted guilt by their actions.

They can say the Passat's system was redesigned because it uses urea. As for the Polo, there's probably a way to spin it as being related to something else, like cost ("Delphi's pump was less expensive for us to buy") or efficiency ("Delphi's pump lets the new Polo use less fuel"). Redesigning something and switching to a different company's product... neither of those are admissions of fault.

Minnesota is Mandated B10 in May 1, 2012.... I wonder how they can sell a TDI in Minnesota?

They're probably going to have to increase the official oil change frequency for MN customers.

 

[emdeeaitch]

They can say the Passat's system was redesigned because it uses urea.

Which could be argued as incidental to the fuel pump, not the same issue. That can be sorted out. The guys at NHTSA are obviously deep into this.

As for the Polo, there's probably a way to spin it as being related to something else, like cost ("Delphi's pump was less expensive for us to buy") or efficiency ("Delphi's pump lets the new Polo use less fuel"). Redesigning something and switching to a different company's product... neither of those are admissions of fault.

The Polo is implicitly an admission of fault, because Delphi/VW claim it is a joint development, and Delphi in their marketing slick bragged that the new design expressly avoids the "fatal" (their word choice) pump damage of the exact sort seen in the Bosch pump. The NHTSA letter specifically calls out VWoA to explain this inconsistency in their position... so put 2+2 together, it is an indirect admission of fault.

 

[koharatx]

I have found a solution to the HPFP potential risk of self-destruction!!!

I'm going to Germany in March, and taking my TDI with me...

 

[specsalot]

It's likely that VW made a business decision to push these cars to market with the unknowns present. Bosch is a storied company, but if memory serves VW really put them on the auto map with the digital fuel injection developed for the 67 VW wagon. VW has defended Bosch's design and also simultaneously walked away. VW has painted itself into a corner with NHTSA. Given they manufacture in the US now, their real appeal will be political and inside the beltway to call off the dogs. All politics aside VW will have significant brand damage with customers over the Bosch pump.

Bosch has no real room to fix things in this design given the failure mode that looks to be in play. Opening up the parts system and promoting periodic inspection / replacement of the follower / roller assembly seems the only logical choice. A simple job for any one with basic skills and an ability to follow directions. The challenge in this approach will be how to handle pumps that show lots of wear in the aluminum bore?

 

[specsalot]

I have found a solution to the HPFP potential risk of self-destruction!!!

I'm going to Germany in March, and taking my TDI with me...

Best of luck with your solution. The core failure appears to be a transition from rolling contact between the dollar and cam to sliding contact. This appears due to loss of anti friction / anti wear coating on the roller carrier that is a light press fit into the follower body. We're all dying to know how you've fixed this aspect of the design.

Folks here are looking at supplementral filtration and potential conversion of the LP body to oil lubrication as partial solutions to prevent full fuel system replacement. Keying the follower is a clear solution to the rotation issue. But it fixes nothing if it creates more swarf in the process of normal operation. Swarf is what kills these pumps pure and simple. The rest of the debate is purely about how they die.

 

[Elfnmagik]

I think he was joking about using better fuel in the Motherland.

 

[specsalot]

Yeah - LOL - I totally missed that. I can't believe the variation in Lubricity is all that significant. There are many places in the world where fuel quality is lower than even US ULSD. Engines seem to run well there. (India comes to mind right away) - HFRR is an excuse for poor design. These failures are design failures first but laid off as fuel spec issues. Canadian and UK failures on eurospec fuel tell us that.

 

[oxford_guy]

There are many places in the world where fuel quality is lower than even US ULSD. Engines seem to run well there. (India comes to mind right away) - HFRR is an excuse for poor design.

India: 2010 Sulfur: 350 ppm (Euro 3; nationwide). 2010 Sulfur: 50 ppm (Euro 4; selected areas)

link

Their fuel still has sulfur, so may it have better lubricity.

 

[kjclow]

link

Their fuel still has sulfur, so may it have better lubricity.

I think it was in the original Spicer report that showed substantial lubricity differences between low sulfur diesel and ultra low sulfur diesel.