Having had a MazdaSpeed3 from 2008 and having been involved with the platform for so many years, one of the most common question that I’ve received – especially in the early days – was why do engines blow on the highway? I went as far as having a collection of broken parts from blown engines all over the country just to see how they failed.

This was a new phenomenon and we saw lots of cars that were bone stock blow up – especially with the first generation MS3 and MS6.

It turns out the Mazda was a pioneer back in 2005-06 when it introduced the DISI – one of the first direct injected spark ignited motors on the market.

Recently (since about 2010-11) the tGDI (turbo gas direct injected) engine has become common so it has received a lot more attention. As more money was spent in downsizing and increasing power output the failure mechanisms in these motors started to become more clear.

And these failures line up exactly with what we are seeing with the DISI.

The root cause is a phenomenon called Low Speed Pre-Ignition (LSPI).

LSPI, which results in Superknock, is auto-ignition before the spark but not early enough to melt the piston. It happens close to the spark event and causes massive knock afterwards. Damaging knock.

The other interesting fact about LSPI is that it happens at low engine speeds and high loads which is a common condition when passing on the highway, for example. Below is a graphic showing the knock in the cylinder just after LSPI occurred:

LSPI is hardest on the rods and pistons in the motor. The rods tend to fail first, and the second ringland follows. This is exactly what we have seen in the MazdaSpeed DISI. Further, most people that lose their motors on the highway or have high knock events on the highway speak about a cloud of black smoke. This is also caused by an LSPI event. From the SAE paper 2011-01-0342:

“From exhaust emission and exhaust port air/fuel ratio measurements it was also recognized that a spike in HC emissions and a significant increase in Lambda (air/fuel ratio enrichment) was associated with LSPI”

Now why does this happen?

There are a number of factors but the biggest are:
– Carbon deposits on pistons and valves, oil fumes, poor quality fuel
– High loads at low RPM
– Valve overlap, EGR

All the causes are still not known. This is still a heavily researched area.

A more complete list below:

The OEMs have learned a lot since the DISI. Take the Ford Ecoboost for example, it doesn’t toss rods like the DISI did. However we have seen it damage pistons and ringlands from the same LSPI phenomenon numerous times. The EcoBoost is an evolution of the DISI using the lessons learned from the early days of tGDI engines.

So what can be done to prevent this condition?

– Don’t operate at high loads and low RPM (Instead of loading the motor at a low RPM, downshift into a lower gear)
– Use quality fuel and lubricants (lots of research going on in this area currently)
– A good tune
– Good intercooling (the TMIC really doesn’t help here on the Mazda), WMI if available
– Good operation of the PCV system, clean intake and valves
– Step colder plugs to prevent hot spots
– Keep oil consumption down – use a thicker oil if needed

The manufacturers have greatly improved piston, injector and combustion design since the DISI. You can even see that Mazda redesigned the pistons in the Gen2 (after 2010) and LSPI was less common on the newer cars because fuel was not washing down the walls of the cylinder as much.

Using common sense can keep the DISI together for many miles, but determining WHY cars blew on the highway has been a very interesting journey for me and it feels like the root cause is finally becoming clearer.

Keep in mind that there is never a single variable that will make a motor come apart and every motor will behave slightly differently. This is good because you can attack different areas of your build to prevent LSPI and superknock. Below is a good summary showing some of the engine parameters and their effect on triggering LSPI.

Not all engines fail due to LSPI. The rods do buckle at certain torque levels and knock can be damaging at higher engine speeds, not just at low speeds. However that strange phenomenon with engines letting go on the highway very much lines up with LSPI being a failure mode in DISI engines.

The images are from a good presentation at the link below. There is a lot of SAE literation on the topic if you’d like to learn more about it.

http://www.ukintpress-conferences.co…_Alewjinse.pdf

Now that we’ve fully completed the installation and optimization of our Water Meth Injection (WMI) system we thought we’d share with you some of the details of the WMI install and tune on our shop Focus ST.

Below is a full list of the modifications we’ve installed so far:
ATP GTX2867 Bolt-on Turbo Kit 0.64 A/R
Catted COBB turbo-back exhaust
Tial Q BPV on CPE METHCharge cold side pipe
CPE Front Mount Intercooler
Stock intake and airbox with COBB filter
Coolingmist Auto-Learn WMI system, 7 Gal/Hr nozzle, 50/50 Water-Methanol mix
92 Octane base fuel

In stock form the biggest power output limitations in the car are airflow, fuel quality, and fuel volume delivered. The stock turbo was swapped out for a GTX2867r on our vehicle to address the airflow problem.

After this the OEM high pressure fuel pump (HPFP) starts to struggle past 360 ft-lbs of torque when using premium pump fuel. The octane limitation of pump fuel itself generally limits power output north of 350hp to the wheels on the 2.0 Ecoboost. To address the octane limitation and fuel pressure drops, the Coolingmist WMI system was used.

WMI has the following advantages:

– It cools down the air charge allowing more ignition timing advance.
– It provides very high octane alcohol (methanol) which increases the fuel octane. The additional alcohol fuel also takes some strain off the OEM fuel system.
– The WMI system helps keep the valves clean in direct injected engines
– The WMI system is only used when needed under high boost meaning the methanol tank won’t need to be constantly filled up.

Adding the WMI system provided around 10% additional fueling with the 7 Gal/Hr nozzle and 50/50 mix. We selected the Coolingmist Auto-learn system which has a built in failsafe. The failsafe system was able to detect our tank running dry as well as a leak that developed at one of the hose fittings.

You can see videos we took of the install here:

Part 1: Under the car

Part 2: Under the hood

Part 3: Inside the car

Results: 

First of all, I should preface that we are very meticulous when taking Virtual Dyno logs to give results that are repeatable and therefore trustworthy. This is very important – you have to trust the data to be able to make sense of it.

The first graph below shows the finalized tune for WMI. The torque curve is very flat from 3000 RPM to redline and the car simply hauls.

Below is a comparison showing the gains we made by tuning the ECU to take advantage of the WMI system.

You can see how the top end benefited the most from the addition of WMI. We were able to safely add timing as well as increase the boost and at the same time keep the pistons nice and happy (no knock). This is what a proper power curve should look like on any car. Our Focus ST now has a flat torque curve with over 300 ft-lbs of torque available at the wheels, from 3,000 RPM right to redline. Notice how linear the horsepower curve is with the maximum power being right near redline.

Now, why no 400whp you ask? There are 2 parts of the car currently holding it back that last 10-15whp. The first is the stock intake which necks down quite a bit just before it joins the coupler on the GTX turbo inlet. The second is the catalytic converter on our COBB downpipe. At 400whp the catalytic converter is easily worth 15-20whp. With these two items swapped out we will be able to get more boost out of the turbo which will take it to 400+whp. The 385-390whp result you see above is done at only 25psi tapering down to 23.5psi. Given that the Focus is used as a daily driver the first upgrade will be the intake. For the intake the plan is to fabricate something that will fit nicely with the GTX turbo inlet. There is something to be said about the OEM intake that  that is that you would never expect this engine bay to be that of a 390whp car.

Finally, I can’t stress enough how much fun it is to have a flat torque curve and perfectly straight horsepower curve. The car is extremely flexible and just doesn’t stop pulling to redline. With relatively minimal mods and a solid tune this little EcoBoost motor really does deliver. A WMI system is very much a worthwhile modification if you don’t have excellent pump fuel and want to move above 350whp on a big turbo. When purchasing a tune from us a WMI kit does not require and additional fuel map so long as you drive with it everyday.

This is a quick observation regarding the effects of replacing the OEM bypass valve with an aftermarket bypass/blow off valve. In general aftermarket BPVs don’t add much performance unless you plan on running 20+ psi on the MazdaSpeed platform. At this boost level pretty much all OEM valves start to open and bleed pressure. On vehicles with a big turbo this can be a little hard to detect as the turbo is able move much more air than the OEM turbo. Big turbo cars will still hold boost (see boost curve on the dyno plot below) but airflow (g/s) values will seem low as will power. On the car below we were in the middle of a pump gas tune on a BNR S3 when the OEM BPV was replaced for an aftermarket unit. The motor had a lot of miles on it and was knock prone but you can see what replacing the OEM BPV did to the vehicle’s performance. This is why we believe it is wise to run an aftermarket bypass valve on tunes where you will be holding boost levels in excess of 20psi as well as all big turbo tunes.

We carry one of the the best BPVs on the market for the MazdaSpeed here.

Aftermarket Vs OEM BPV