Professional Documents
Culture Documents
The aim of this article is to give a clearer understanding about the viscosity of engine oils.
First off a definition of viscosity: it is basically the resistance to flow of a liquid due to internal
friction. You see this when you pour honey – it is very viscous or ’thick’. Water is much
‘runnier’, oil is somewhere in between. The more resistance to flow a liquid has, the higher the
viscosity.
Let’s move to the motor in your car. Viscosity of the oil is its most important feature. The oil
needs to be viscous enough to prevent sliding surfaces (crank bearings, cam lobes, even
piston rings in bores) from rubbing on each other by maintaining a thin film between those
surfaces. The viscosity needs to be high enough to prevent the oil from being squeezed out
faster than it can be supplied by the oil pump (or splash feed or whatever). One way of
monitoring this in your own car is by observing the oil pressure gauge – if the oil doesn’t have
enough viscosity for the available pump flow and bearing clearances, etc, then the oil will
basically leak out of the bearings too fast and the oil film will become too thin, resulting in
nasty metal to metal contact. Of course you will be saying that maybe the problem is really
your worn out bearings or dodgy oil pump. Sure, they may be out of spec – you either replace
them (best solution) or use a more viscous oil to get by.
Imagine that you want to lubricate a main crankshaft bearing in your motor, one metal surface
is stationary – the bearing itself – and one surface is at a high speed – the crankshaft journal.
The oil immediately in contact with each surface (in the first few nanometres) is moving at the
same speed as that surface. Through the oil film between the two surfaces the oil has a
velocity profile that increases from zero at the bearing surface to a maximum at the crank
journal. This means there is shearing of the oil film – sort of layers of oil sliding over each
other. The higher the viscosity (internal friction remember?) the more oil drag there will be.
This means there is resistance to flow so the oil doesn’t immediately leak out the ends of the
bearing and there will be a build-up of pressure in the oil galleries. Consider that the oil pump
doesn’t generate pressure itself. It only moves oil from its inlet to its outlet. There is only
pressure generated when there is a resistance to that flow. Other effects of the oil shearing in
the bearing are that there is some drag trying to slow the crankshaft (relevant for fuel
economy) and there will be some heat generated.
Another big function of the oil is to provide a seal between the piston rings and the bore (as
well as lubricating these sliding surfaces). Viscosity is again relevant.
There are two ways of specifying viscosity – kinematic and dynamic. Frankly I don’t know why
they bother because dynamic viscosity is just the kinematic viscosity multiplied by the density
of the oil (which you can always take as 0.86 grams per cc, so it is a constant anyway).
Throughout this article I have standardised on dynamic viscosity. It has units of centi-poise
(cP) where 1cP = 1 milli-Pascal second (mPa.s).
So, in order to satisfy the requirements of maintaining oil films in bearings, sealing piston
rings, lubricating cam lobes, etc - taking into consideration the bearing and cam sizes and
loadings all built in by the motor designer - and not wanting to waste fuel or generate
excessive temperatures, then there must be an optimum viscosity for the oil. Maybe its not
too critical – so the viscosity can be in some range, say plus or minus 20%. No problem you
say, they have all different oil viscosities in the shop, so it is just a matter of picking the one
that suits my motor. Unfortunately there is a huge problem. Any given sample of oil changes
its viscosity with temperature. MASSIVELY. Have a look at the graph below showing viscosity
change with temperature for an “SAE 30” engine oil. (More on the “SAE 30” later).
Absolute Viscosities - Logarithmic scale
100000
Changes a bit doesn’t it? Look
closer, I’m using a
compressed (logarithmic)
scale for the viscosity. At -
10000
20ºC (relevant in Canada,
USA, Northern Europe, etc)
Viscosity (centiPoise)
1
-40 -20 0 20 40 60 80 100 120 140
Temperature (degrees C)
So, what the SAE (Society of Automotive Engineers) in the USA have done is the next best
thing. They had a ‘runny’ oil in a bottle on the shelf that they declared was “SAE 0 motor oil”
and a ‘thick’ one that they labelled “SAE 60 motor oil”. Then they made blends of these oils to
come up with intermediate grades like SAE 20, SAE 30, etc.
They all had curves like the one at left but they just moved the curves up (more viscous,
higher SAE number) or down (lower viscosity, lower SAE number).
To be practical, they made up a list of criteria for oil producers to use so they could rate their
own oil. This specifies the maximum viscosity an oil is allowed to have at a very low
temperature (about -20ºC) and the viscosity it must have at 100ºC. Over the years, these
specs have changed slightly and even some extra viscosity tests added. To avoid the
(pointless) manufacture of precision oil (after all, viscosity changes SO much with temperature
anyway), any given SAE grade covers a range of allowable viscosities.
Here is the SAE table released in April 1997. (It has been superseded by one in Dec 1999 but
it just changes the way the low temp figures are presented and is not as clear for our
purposes) :
SAE Low Temp Viscosity High Temp Viscosity High shear at 150C
Viscosity Maximum (cP) @ temp (C) at 100C
Grade Min. (cP) Max. (cP) Min. (cP)
0W 3250 -30 3.3
5W 3500 -25 3.3
10W 3500 -20 3.5
15W 3500 -15 4.8
20W 4500 -10 4.8
25W 6000 -5 8.0
20 4.8 8.0 2.2
30 8.0 10.8 2.5
40 10.8 14.0 2.5 (0W40, 5W40, 10W40)
40 10.8 14.0 3.2 (15W40, 20W40, 25W40, 40)
50 14.0 18.8 3.2
60 18.8 22.4 3.2
Notice in the above table, due to the allowable range for any grade, that one company’s 40 oil
may be practically the same as another’s 50 oil. Or another way of looking at it: one brand’s
30 oil may be nearly twice as viscous as another brand’s 30 oil – or it may vary between
batches or over the years from the one manufacturer.
So far, I have only discussed ‘straight’ grades of oil. In order to flatten out some of these
horrible viscosity-temperature curves, oil producers have added a Viscosity Index (VI)
Improver to a lower viscosity grade (like SAE 20). These are polymer molecules (like plastic)
that have the effect of making the oil more viscous than normal as temperatures increase. So
what you end up with is an oil that has the viscosity of (say) an SAE 20 oil at very low
temperatures but has the viscosity of (say) an SAE 50 oil when it is at 100ºC. They then call
the oil “SAE 20W50”, ie the oil is a 20 Winter grade (for starting in a North American winter)
and a 50 grade when fully up to temp in a hot climate.
I have tried to mathematically model these curves from available data (internet and books).
This worked reasonably well and is shown in the table below and in the graph on the next
page.
10000
Viscosity (centiPoise)
5
1000 10
20
30
100 40
50
5W50
10
1
-40 -20 0 20 40 60 80 100 120 140
Temperature (degrees C)
See how the multi-grade oil (5W50 in this case) has a flatter curve. It has a more consistent
viscosity. This is the curve that starts the lowest at the left and ends the highest at the right.
Notice how the viscosity of the 5W50 doesn’t drop off as much as straight 50 when the
temperature goes over 100ºC.
So is there any reason why you shouldn’t be using multi-grade oil with the biggest spread of
SAE grade numbers (like 0W60 if it were available)? And why have you seen oils marketed as
racing oils that are straight 30 grade?
Well it seems that, at least in the early versions of multigrade oils, the VI improvers that were
used (maybe in just some brands?) were not particularly high temperature stable over time.
This meant that in your racing engine (that therefore ran high oil temperatures) the
temperature would ‘cook’ the VI improvers and you would end up with the 20W50 oil reverting
to straight SAE 20 oil with some gunge in it. The resulting low viscosities would of course stuff
the motor. Apparently modern VI improvers are much better.
From what I can tell, there is no longer any justification for those straight grades of ‘racing’ oil.
They are probably marketed towards people ‘stuck in their old ways’. Those people are happy
to stick to their pattern of extra careful engine warming, frequent oil changes to get rid of
those wear particles and/or routine bearing changes “because that’s what you do with a
race/performance car”.
The other issue is the tendency for VI improvers (which are long tangly molecules) to
straighten out and become less effective when placed under high stress –like under a cam lobe
(and maybe a crank bearing?). Once out of the high stress area the VI improvers get tangly
again so the oil fully recovers. How much of an issue this is I don’t know, as oil viscosity also
naturally increases when under pressure. (If this added variable starts to blow your mind,
then just forget I mentioned it).
Anyway, you can leave this whole VI improver issue behind (or at least minimise it) by going
to a synthetic oil. This is because the flatter viscosity characteristics are ‘built in’ to the design
of the oil molecules when they are being synthesized. Some info I have from maybe 10 years
ago says that some synthetic oils use a small amount of VI improvers to extend their range.
How this applies to what you can buy today, I don’t know.
Conclusion: Buy 5W60 or 5W50 fully synthetic (whichever suits your motor better).
A Better Conclusion: You’re going to need a helluva lot more research and information to
select the right oil.
Good Luck.