What can Used Oil Analysis tell us about a failed engine?
update
Cover Image for What can Used Oil Analysis tell us about a failed engine? | 900BRZ

Background

After my engine failure earlier this year while testing a prototype baffle, I sent a sample to Blackstone Labs with no indication of what had happened to see what they would detect in the oil.

Some people I respect have told me that they think about Used Oil Analysis (UOA) as useful for determining the properties of engine oil, but less useful for determining the health of the engine. They argue that the primary benefit is understanding things like whether the oil is shearing down to a lower viscosity, which may cause you to choose a differenet oil.

There are certain engine failure modes that should be relatively easy to detect with UOA, like a head gasket leak causing coolant to enter the oil. But the type of engine health we’re most interested in for Subaru engines is bearing health, which requires detecting trace metals in suspended in the oil.

So, I decided I would send in the sample from my failed engine even though the health of the engine was not in question. Instead, I hoped to test my friends’ theory that UOA is not useful for detecting bearing wear. I would also hopefully pinpoint exactly what a bad UOA looks like for an FA24. For example, which metals should we be looking for to indicate bearing wear?

Results

Here’s the report comments from Blackstone:

This oil wasn’t in use very long at all, so we're surprised to find wear metals around the same levels as last time (if not higher). We also found some visible ferrous debris (steel) in this sample, which, combined with these results, could show an issue. Or maybe this run included hard use and the engine left more metal behind as a result. In any case, it's best to take a cautionary approach. If the engine runs well, then let's see how metals look after another oil run. Hopefully iron shapes up on a microscopic level and visible metal disappears. No contamination was found.

Overall, the comments show concern, but the UOA is still ambiguous, especially compared to the rather obvious sight of glitter in the oil pan.

When you look at the results table in the report, metal levels are high, but not significantly higher than some of the previous samples. The largest deviation is 8 ppm of copper, compared to 2 in the previous sample. But the universal average for copper is 7 ppm and I had a 5 ppm unit average for this engine. I was expecting to see an order of magnitude difference in at least some of the metals.

Blackstone UOA report. Leftmost column is the sample from the failed engine. From left to right, oils used are Motul 8100 EFE 5W30, Motul 8100 EFE 5W30, Motul 8100 EFE 5W30, Motul 300V 5W30, Subaru 0W20.
Blackstone UOA report. Leftmost column is the sample from the failed engine. From left to right, oils used are Motul 8100 EFE 5W30, Motul 8100 EFE 5W30, Motul 8100 EFE 5W30, Motul 300V 5W30, Subaru 0W20.

I sent email to Blackstone describing the failure and asked the following questions:

I was expecting to see higher metal traces, as I was thinking this would give us an indication of which metals to look for to indicate bearing wear in the FA24 engine. Should we not expect UOA to be a good indicator of bearing wear? While we see elevated levels of iron and copper, it's not massively more, as I expected. For example, copper is only 30% higher than a previous test and iron is only 10% higher than unit average.

I quickly received a response from a Blackstone representative, Samir Kharbas. I’ll include the bulk of his response because I think it’s helpful for people to understand:

Engine failures don't always show up in analysis when they're sudden failures. In this case, the baffle was blocking the oil pick up screen and that resulted in rod bearing failure, but maybe not excess bearing wear, if that makes sense. The bearings weren't wearing poorly and leaving excess metal behind from sample to sample, which is what we find from an engine that's developing a bearing problem.

It sounds like this might be a case of a spun bearing, which happens suddenly, not over time. The sample also had visible ferrous metal (steel), which is probably from the failure, but visible metal is too big for our machines so that metal isn't reflected in the results.

Bearing wear for these engines usually shows up as aluminum. In fact, that's true for most modern automotive engines because manufacturers are trying to move away from lead due to toxicity. We also find copper and tin (bronze - second layer) from aluminum bearings.

It’s hard to gauge wear levels in this sample against the previous samples, mainly because the oil wasn't in use as long. If you look at metals on a per-mile basis (ppm/mile), metals are certainly higher than before. We give some leeway with short oil change intervals because a lot of the metal we're seeing is residual from the previous fill. It’s almost impossible to get all the old oil out in one go.

We do a lot of insurance samples where trends simply aren't an option, and we usually find a lot of metal when engines fail, but if the engine was healthy and then failed suddenly, then the evidence may not be in the used oil.

Our analysis includes information about the oil's physical properties and when appropriate, we discuss how the physical properties may be influencing wear. The physical properties for this sample were excellent.

He also provided some further clarrification on the size of metal particles that can be detected with UOA:

I’m not 100% sure on the smallest amount of metal we can detect, but I want to say it’s somewhere in the ~5 micron range (or smaller). You are correct that anything visible with the naked eye won’t be reflected in the analysis. Something like "glittery" oil, for example, won’t always result in high amounts of microscopic metal. It isn't unusual to find extra metal on a microscopic level if visible metal is present, but depending on what’s causing the metal, we may or may not find extra metal in analysis.

Conclusions

First I want to say thank you to Blackstone for providing detailed and helpful responses to my questions. I appreciated that they were open with some of the limitations of UOA.

They point out that the short change interval (a single track day and a few hundred miles total) makes it difficult to compare the results to previous samples. And they suggest looking at ppm/mile, which is an interesting idea to try and normalize the results for different change intervals.

The fact that UOA analysis can’t detect visible metal particles is not particularly concerning to me. If the metals are visible to the naked eye, we should use our eyes!

It’s a good idea to change your oil yourself and ideally use an open catch container, so you can inspect the oil for any debris before disposal.

UOA is an important tool in our toolkit, but it’s useful to develop an intuition around what its limitations are so you don’t take away the wrong conclusions from, for example, a clean UOA on a used car. Just because the analysis is clean doesn’t necssarily mean the engine is healthy.