At the end of last year, my wife and I took a short break in France. Worried by the UK Government’s Renewable Transport Fuel Obligations and next year’s planned increase of ethanol in petrol to 10% (E10)-but see below- I took the opportunity to bring back 10 litres of “supermarche” E10 to evaluate the damage it might do to my TC. As a test, I immersed samples of materials, representative of those found in our fuel system, in small containers filled with E10 and stored them in my garage over winter. What I found was not what I expected.
Fortunately, in February this year the Department for Transport (DfT) invited Federation of British Historic Vehicle (FBHVC) member clubs to a Fuel Stakeholder meeting. One of the key messages which DfT were keen to put across was that E10 is definitely not going to be introduced in 2013. However, the problem has not gone away, the introduction of E10 has only been postponed and worryingly, E5 (5% ethanol blended fuel) is with us now.
To confirm that any damage to my samples was due only to ethanol, I used three “fuels”, BP Ultimate (premium 97 RON) as a control (which is ethanol free in our area), E10 with added water to “wash out” the ethanol (see the section on Contamination) and “pure” E10. I placed samples of rubber, cork, fibre washer and leather retrieved from my stock of old TC bits and different pairs of electrically connected steel, copper, brass and aluminium plates in separate small pots filled with these different fuels (see photos 1 and 2). I then left these samples to “stew” in my garage for 4 months over winter alongside the TC. In this way they were subject to the same conditions as the parts of my car to give a representative assessment of any damage.
The FHBVC has identified three areas of concern with the use of ethanol blended fuels; Compatibility, Corrosion and Combustion. My tests have added a fourth – Contamination. I apologise, in advance as my examples apply to TCs, however, many of the things I describe are common to all pre-mid 1950s MGs and most later ones as well.
So how did my samples stand up?
Compatibility: Put simply, ethanol rots anything that is non-metallic and this is the area that is written about the most. While there are various lists of materials at risk, the main concerns for us are the cork gaskets in the 2 1⁄2 gallon sender unit in the tank, cork seals in the carburettor jets, fuel pump diaphragm and two hoses connecting the pump to the carburettor and possibly the fibre washers on the float chamber lids. Originally, all the washers on the unions are copper so these are not included in this category. Early, non-ethanol resistant, slosh tank sealants are also at risk.
Compatibility appears not to be something to worry us. Firstly, none of my samples showed any degradation, including the cork gaskets (note: cork is listed as ethanol safe by some articles). The only problems I detected were with the Perspex spacers I used with the pieces of metal and with a silicon sealant used to seal the lid on an old float chamber. The latter is a concern if you have used modern silicon based gasket sealant in your carburettors as this was affected just by the vapour.
My advice is to listen to the ticking of your electric fuel pump when you switch on the ignition. It should start “enthusiastically” as it fills the float chambers then stops- dead. Slow initial ticking and you may have partially blocked fuel pipes due to the rubber swelling or blocked filters. If the clicking does not stop you may have leakages in your fuel pump diaphragm, valves or hoses. The bottom line is that compatibility issues are easy to detect and simple and cheap to fix. However, do watch if your tank has been slosh sealed for softening or blocked filters in the tank or fuel pump.
Contamination: In particular by water. Ethanol absorbs water from the atmosphere and once a certain level is reached it can “phase separate” with an alcohol/water mixture settling at the bottom of the tank with the remaining hydrocarbons sitting on top. While it is unlikely that absorption of moisture from the air on its own will cause this, a sudden temperature change or even a drop of rainwater entering the tank can result in phase separation. If a petrol tank filled with 5 gallons of E10, undergoes phase separation, you end up with 1⁄2 gallon of alcohol/water and 4 1⁄2 gallons of petrol, no matter how little extra water enters the tank. Worse still, you will probably not know this has happened, as an engine will run weak on the pure alcohol and as this is used as an octane enhancer it will probably pink as it burns the remaining fuel.
The second effect is of more concern. As the fuel sloshes about in the tank or is stored, droplets of alcohol/water condense above the fuel (Photo 3). This will result in rusting of an unprotected fuel tank and is probably the cause of the rust spots reportedly seen in tanks used to store ethanol blended fuel.
Practically, this is more of a worry than Compatibility and you do not have to wait for E10 to arrive for it to become a problem. It is with us now. Photo 3 was taken with current UK supermarket petrol, probably E5.
My suggestion is to get your fuel tank slosh sealed with an ethanol proof sealant to protect it from rusting and avoid storing large quantities of ethanol blended fuel in your tank – so even if it does phase separate, there will only be a small quantity of fuel present to damage your engine.
Corrosion: What compatibility does not get, corrosion will, anything metallic! This shows itself in two ways; oxidation and galvanic corrosion between different kinds of electrically connected metals. The steel, brass, copper aluminium metal plates were used to test both these corrosion mechanisms. To maximise the probability of any oxidation, the plates were abraded, cleaned and degreased prior to storing in the fuel. The different pairs of metals were spaced about 1/8” apart and electrically connected to assess the effects of galvanic corrosion. This gave 18 pairs of plates, 6 for each of the different fuels that were being tested. The choice of metals was to represent those found in the fuel system of a TC, sorry TA owners, no bronze.
There are many places in the fuel system where these different metals are in sufficient contact to allow an electrical current to flow. The fuel sender is aluminium bolted with steel bolts into the steel tank, the fuel pump has an aluminium body with brass valves and the float chambers are aluminium with a bonded steel rod to name a few examples. I also tested an old float chamber by sealing off the holes and filling with pure E10. This is still under test.
The FBHVC has also been running oxidation tests with polished steel probes to evaluate the effectiveness of additives. Unfortunately, these have been plagued by contamination problems delaying the results. In my tests, with the exception of a very small area of minor corrosion on a steel plate in the E10/water mix, where the ethanol had phase separated, none of the other samples were affected by corrosion. This suggests that the French E10 already contained an inhibitor that prevented any oxidation. Additionally, if your tank has been slosh sealed the only steel part at risk is the fastening bolt in the float chamber and this is unlikely to be damaged by minor corrosion. My conclusion is that oxidation is not a worry and there is no need to spend money on additives.
The story with the galvanic corrosion was somewhat different. Before electrically connecting the plates, I measured the potential difference between them. As may be expected, in the BP Ultimate 97 RON (ethanol free) and E10/water samples, I could not detect any voltage. However, in the E10, I did detect a potential difference between the plates. This is direct evidence of galvanic action. At the end of the test, the potential had risen significantly suggesting the corrosion affect had become worse over time (see table 1). The most affected samples were aluminium, particularly when connected to steel. The aluminium plate from the steel – aluminium pair was visibly marked with numerous corrosion pits. Photo 4 shows the aluminium plates paired with steel from the BP Ultimate (no ethanol) and E10 clearly showing the classic corrosion pits.
Worrying, as these have resulted from only 4 months immersion in E10.
Table 1 – Measured voltages (mV) – E10 only
Galvanic corrosion between steel and aluminium is a real issue. Unfortunately, the use of die cast aluminium with steel parts is common in the fuel systems of most vehicles, modern cars included. Furthermore, stainless steel appears to be worse than mild steel giving a greater potential difference, leading to a higher rate of corrosion. What is more worrying is that I measured a significant voltage between steel and aluminium plates using our own UK supermarket fuel which probably contains 5% ethanol. The aluminium parts on your car may be slowly dissolving even as you read this article.
In pre 1956 MGs the main parts at risk are the fuel sender, aluminium in a steel tank and the float chamber with its steel bolt and steel union bolts. Fortunately, unlike modern cars whose aluminium bodied high pressure fuel pumps are manufactured to very tight tolerances, the main risk to these parts is that they become porous. It is possible to avoid this problem when storing your car by ensuring there is less than 2 1⁄2 gallons of fuel in the tank (i.e. it is below the level of the sender) and installing a switch to switch off the electric fuel pump. Rather than switching off the ignition to stop the car, switching off the fuel pump and this will empty the float chambers before the engine stops.
While there was observable corrosion to the aluminium when paired with brass and copper, there were around half the number of corrosion pits in the aluminium than with the steel sample.
The parts less at risk are the fuel pump with its brass valves in a die cast aluminium body and the carburettor body with its brass jet.
These tests suggest galvanic corrosion is a far more serious issue than oxidation and not only does it pose a serious threat to all classic and many modern vehicles, there is very little that can be done to mitigate this effect. Checking for the possibility of galvanic corrosion is relatively easy, place two samples of different metals in ethanol blended fuel in a non-metallic container, ensure they are not touching each other and measure the voltage between them. If a voltage is present, they will corrode. Interestingly, when I tested my samples with a 10% kerosene / E10 mixture, the measured voltages dropped by 40% suggesting that adding kerosene may be a route to reducing the rate of galvanic corrosion. Certainly more research in this area is needed.
Combustion: There are a number of differences between ethanol blended and modern fuels in general and the classic fuels our cars were designed to run on.
Enleanment: This is a problem with ethanol blended fuels. As the alcohol contains oxygen, less air is needed to burn the mixture. Running with the carburettor set for normal fuel will result in the engine running lean risking overheating and valve damage. This is a simple problem to detect and fix. Your plugs should be a biscuit brown colour and with an SU carburettor the mixture can easily be enriched by screwing the mixture adjusting screws down by 1 or 2 flats.
Specific gravity: A problem with all modern fuels is they have a higher specific gravity than classic fuels. This causes the floats in the float chamber to “try” to float higher, depressing the fuel level in float chamber and jets. This, in turn, leads to a poor air fuel mixture distribution in the carburettor, poor starting and rougher running. Set the forks in your float chambers using a 3/8 bar as per the manual, remove the dashpot and then add weights to the float until the level of the fuel in the jet is 3/16” below the top of the jet. To do this, I pull out the choke to drop the level of the jet by 3/16” then set the fuel level to the point it just overtops the jet. Additionally, a higher specific gravity fuel enriches the mixture which again with SU carburettors is easy to adjust for. Unfortunately, there is no relationship between ethanol content and specific gravity so these effects do not cancel each other out.
Vapour pressure: All fuels consist of a range of different components with different boiling points. The vapour pressure at a given temperature is, in effect, a measure of the percentage of components that will boil at or below that temperature. To improve starting, modern fuels have a higher vapour pressure than classic fuels. The FHBVC suggest, in a running engine, this causes vaporisation in the fuel system which in turn leads to a weak mixture and hot running. They advise fuel system components should be insulated to avoid this problem. While the higher vapour pressure is certainly a cause of difficulty in restarting an engine when it has been stopped for a few minutes after a run, I believe this vaporisation is only a symptom of a more fundamental problem that causes hot running, namely slow burning.
Slow burning: The evidence for this has come from two sources. A few years ago I had my car tuned on a rolling road and the only change was to advance the ignition to 11o advance at tickover. After rebuilding my distributor, the acceleration and exhaust temperature tests run by David Heath and myself, “Modern Fuel on Trial” (see TTT 2- Issue 2 –October 2010), showed best performance occurred with the ignition set to 13o advance at tickover. Compared to the original setting of 4o advance at tickover it shows that modern fuel is taking longer to burn and reach peak pressure in our engines. The speed at which the fuel burns depends on the turbulence of the gasses in the cylinder. Unfortunately, in each cylinder on each stroke, minor variations in the initial conditions can have a major effect on the time it takes for the gasses to fully burn and reach peak pressure. This effect is called cyclic variability and shows itself as rougher running. The slower burn of modern fuels gives more time for the amplitude of this effect to increase. Cyclic variability can be thought of as “smearing out” of the ignition timing, for example with my engine running at 2000 rpm with 25o advance the actual timing may vary between 20o to 30o advance due to cyclic variability. Obviously as the amplitude of this effect increases, the risk of some cycles causing pinking or the fuel still burning when the exhaust valve opens is increased. My recommendation is that you ensure there is as little slop in your distributor as possible and the centrifugal advance is working properly so at least wear in the distributor is not adding to this effect. I also recommend you ensure your timing is set as accurately as possible, either by visiting a rolling road or by measuring acceleration as David and I did.
I also have made measurements that show there is a significant difference in the way different brands and grades of fuel burn. Analysing these measurements is complex and is still on-going. Because petrol from the same distributor differs when and where you buy it, I suggest you try different brands, including the super grades, try adding up to 10% kerosene and choose the one your car runs smoothest on, i.e. with the lowest cyclic variability.
I have found my standard tune TC runs best on an ethanol free super-grade fuel with 10% of heating oil, returning a very healthy 35mpg on a run, suggesting the engine is in a good state of tune. My understanding is that in the UK “Super” (95 RON) is likely to contain up to 5% ethanol and “Premium” (97 RON) is probably ethanol free. However, there will be exceptions in certain parts of the UK due to distribution logistics.
With no added insulation or a heat shield, my car does not run hot and I have not suffered from vaporisation problems since I rebuilt my distributor and advanced the ignition. Further evidence that the lower vapour pressure of modern fuel is not the root cause of this problem.
So what of ethanol blended fuels? My conclusion is that with care our pre 1956 classic cars are probably better able to cope than more modern vehicles. The main area of concern appears to be galvanic corrosion of aluminium and while compatibility may still be a longer term problem, it is easy to detect and cheap to fix. For me, combustion is still a worry because of the risk of expensive engine damage and my car certainly runs better on ethanol free fuels. John Saunders has fitted an MGB distributor with a vacuum advance to his TC which improves the timing accuracy. He reports significant improvements. My next steps are to investigate programmable distributors (123ignition.nl) which will both improve timing accuracy and allow me to map both the centrifugal and vacuum advance curves. Watch this space!