Improving the 2.0 liter Ecotec DI-VVT Turbo (LNF)

[dwightlooi]

The LNF is a very good engine. 260hp @ 5300 rpm and 260 lb-ft @ 2500 rpm are very good numbers. BUT, they are not as good as they can be. Here's why...

This engine revs to 6300 rpm, but the last 800 rpm or so does nothing but give the driver the flexibility of not shifting if he is in a corner and does not want drive train disruption at that specific time. Other than that, this is one engine that should be short shifted way shy of its redline. The KKK K04 turbo used is again, good but not the best. This is very similar to the unit used in the Audi TT 20v 1.8T (225hp version) and is a little undersized for 260 hp. The response of the engine is good, but not as good as some lower boost turbocharged engines like the 2.0T FSI from VW-Audi group. It is efficient for its output, but again not stellar in this department.

How can we make it better? Well, I think that it is possible to improve engine responsiveness, push about 300 hp from the engine, make the engine enjoyable all the way to the 6300 rpm redline and improve economy. Here's how...

Change #1: Decrease Boost

This may sound like a retrograde step, but it really is not. The reason is three fold. First of all, it allows us to increase the compression ratio of the engine which improves light load response and make the engine feel more "alive". Secondly, when I went through all the compressor maps of most of the turbos from IHI, Mitsubishi, KKK and Honeywell (Garrett), there doesn't appear to be one which has their broadest efficiency bands at 1.25~1.35 bar (18~20 psi) which is what the current LNF is running. I believe that this is the limitation of a single stage centrifugal impeller. There are however quite a good selection with maps that are down right fantastic maps at ~1 bar (14.7 psi). This is important if we want a big flat torque plateau across a wide rpm range. 3~5 psi is not a lot of boost difference and we can hit 260 hp at 14.7 psi on a 2 liter anyway, so bear with me a little.

The following are the compressor maps from a Garrett GT2560R and from the KKK K04. Both units have better efficiency working at around 1 bar than at 1.3 bar. The GT2560R is so good in fact that efficiency reaches 78% and never goes below 60%, the K04 reaches 72% and maps as low as 55%.

Lastly, for any given compressor and turbine wheel efficiency and inertial, it takes a shorter while to reach 15 psi than it does 18~20. This means an improvement in lag time between off boost lugging and full boost scooting and we all know that's nice.

Change #2: Increase compression

Now that we have backed off on boost a little we need to make up for it by upping compression a little. How much? How about one full point to 10.2:1 (the LNF is 9.2:1). This is about right. For instance the VW-Audi 2.0T FSI runs about 12.5 psi on 10.5:1 with a K03 turbo.

This does three things for us. It makes the engine more responsive off boost. It makes it more economical on the freeway and in gentle driving -- most of which will be done with manifold pressures in some degree of vacuum; -0.4 ~ 0 BAR. It also recovers some of the power and torque lost through the decrease in maximum boost.

Change #3: Reduce the pressurized volume

The big front mount IC on the LNF is very efficient at dropping charge air temperature. But, it also creates a big volume of air to be pressurized. Think of the turbo as a compressor pump. If you an electric pump to pressurize a basket ball to 15 psi it takes a very short time. Use it to pressurize a sealed room and it takes forever. Basically, big ICs and long hose routings decreases response and increases boost lag. So, ideally we want the pressurized volume to be as small as possible while still meeting our desired charge cooling targets.

Now, having decreased the boost a little and working off a more efficient part of the compressor map helps by not heating the air as much and hence reducing our charge air cooling demands. But we can do more. Let's dump the air-to-air intercooler and adopt an air-to-water unit. Water is a much better carrier of heat and the heat exchanger can be as small as a brick and be as good as that big front mount IC. Its size also allows us to mount it on the cylinder head. Basically, the air leaves the turbo, goes through this tiny air-water exchanger placed near the valve cover and go straight to the intake manifold. The pressurized volume is probably about 1/5th that of the current setup (if not smaller).

The air-water solution of course still needs a radiator to be mounted somehwere, possibly where the current front mount IC is, but the distance and size of this radiator will not affect pressurized air volume.

Change #4: Increase the stroke

Normally, I am not a fan of stroker motors for a variety of reasons. But in this case I believe it is warranted.

The reason is that a 6300 rpm red line doesn't need a 86mm stroke. We can run a longer stroke and still be well within the piston speed limits. Despite what some people may think, power and torque doesn't limit an engine's redline much. Piston speed does. The reason is that the stressload on the rods and journals increases linearly with torque increases, but exponentially with rpms. Why? Because you are slowing and accelerating piston slugs and the kinetic energy you need to slow from and accelerate to is a function of the square of velocity.

Also, increasing stroke length increases displacement, but it DOES NOT increase the combustion chamber size where in matters (near TDC) when ignition events occur. Hence, it does not degrade knock resistance of the engine. Having a slightly undersquare bore x stroke ratio is also ideal for allowing us to extract more energy from each fuel/air charge, while still maintaining a good valve area for the given displacement. Just about all the reallly good turbocharged engines like the Mitsu 4G63 and the VW-Audi 1.8/2.0Ts are undersquare. The Subarus are not, but that is because its a boxer and they can't make it any wider! This is also partly why the Subaru WRX STis have 8.0:1 compression whereas the Lancer Evos run 8.8:1.

For family commonality, let's simply run the stroke length of the 2.3 liter Ecotec motor (90mm). At 86 x 90 mm, this will yield a 2.1 liter displacement.

Change #5: Use a GT2560R ball bearing turbo.

The K04 is a journal bearing unit, a ball bearing turbo spools faster and is arguably more durable when the oil properties are less than ideal. The GT2560R is very compatible with the airflow requirements of our 300hp target and has a peak compressor efficiency of 78% and peak turbine efficiency of 75%. This is about 6% and 10% better than a K04.

Of course the twin scroll manifold design should be maintained. Dual scrolls do not actually direct exhaust onto the turbine better as some people believe (it is actually a little worse due to increased wall drag on the airflow). However, it prevents parasitic exhaust pulses from reaching the cylinder in its intake-exhaust valve overlap period while the cylinder on its exhaust period is exhaling. This reduces the contamination of the engine's breathing cycles making it more efficient and also prevents the loss of pressure that is needed by the turbine from being partially lost to cylinders on the intake phase.

Change #6: Use a variation of AFM (aka DoD) to allow part-time Miller Cycle operation

Now this is a little complicated so bear with me...

An engine that keeps the intake valves open notably into the compression stroke is sometimes called an Atkinson Cycle or Miller Cycle engine (the differences between the two are in aspiration assist methods).

Typically, a turbocharged engine benefits from the late closure of the intake valves. This is because with the pressurized intake air, the engine can push air into the cylinders somewhat into the compression stroke even if piston is going up! In fact, this is desirable because the restrictions from the valve area being smaller than the cylinder bore (which is always the case) means that at BDC the cylinder is not completely filled to the same pressure level as the intake manifold. In normally aspirated engines the cylinders are sucking air into themselves through vacuum action hence as the cylinder is going up, there is very little ability to do so. At very high rpms, they are able to do it somewhat from the supercharging effect of the closing valves building up a temporal positive pressure behind them as high speed airflow gets suddenly stopped and air stacks up behind the valves. But that is another topic for another day. The key issue here is that turbocharged engines have very considerable ability to aspirate into the cylinders somewhat into the compression stroke whenever there is boost present regardless of engine speed. The same goes for supercharged engines.

However, a turbocharged engine does not always make boost, and if we keep the valves open into the compression stroke it will decrease the engine's output when off boost. It may also negatively affect emissions because we have effectively decreased compression ratio by "kicking" some of the intake charge back out the cylinder (sometimes with fuel already in there) as the piston goes up.

Ideally, we'll use a VTEC or VVTL-i style cam switching system to switch between our regular (Otto) cycle operation and Miller cycle operation. But that adds a whole different level of complexity and cost to the cylinder head design. AFM -- Active Fuel Management or Displacement on Demand -- has been used successfully in many GM engines. It has not been employed in 4-potters such as the Ecotec family. But with some additional passages in the heads there is no reason why it can't.

What I am proposing is not a fuel economy idea, but one for performance. We will incorporate AFM onto one of the two intake valves for each cylinders. One of the valves follows and Otto Cycle cam, whereas the other follows the Miller Cycle cam. Off boost and at idle, AFM collapses the lifter on the Miller Cycle valve and it never opens. The cylinder is fed by the Otto valve only and closes the intake valves early. The engine also benefits from increased swirling of the intake charge with one intake valve. Once we develop a reasonable amount of boost (say ~5 psi), AFM solidifies the lifters and opens the Miller Cycle valve. The Otto valve opens and closes as it used to, but even after it closes, the second valve remains open feeding the cylinders with compressed air somewhat into the compression stroke. This increases volumetric efficiency and in also creates an asymetrical compression and expansion stroke which is desirable for extraction more energy from each drop of fuel (this is why the Prius uses an Atkinson Cycle engine even though it is NA and reduces the power yield per liter).

The concept is simpler than say VTEC style cam switching and the key is that the engine initiates the Miller Cycle mode operation on boost.

End result

This about it! Conservatively, this should yield a 2.1 liter engine with about 260hp @ 2200~6200 rpm with about 310 hp @ 6300rpm and redlining at the same 6300 rpm. On top of that, we should have made it more responsive, more economical and made it desirable to rev all the way to the red line. Essentially, we have taken the LNF torque flat and broadened it to a higher rpm without compromising the engine speed at which boost first hits.

[ZL-1]

I'm no engineer, but why not double turbocharging? smaller turbo for low revs, two turbos for mid-range, and bigger turbo for the upper-rev range.

[thegriffon]

There is no 2.3 L Ecotec motor. You must be thinking of the Saab 2.3 L, which is completely unrelated to any other GM engine family and uses a wider bore spacing (and hence a longer crank). The Saab Family 3 4-cylinder is a pre-GM block dating back to 1972, redesigned in 1984, and originally based on the smaller Triumph slant 4 previously used by Saab.

[dwightlooi]

I'm no engineer, but why not double turbocharging? smaller turbo for low revs, two turbos for mid-range, and bigger turbo for the upper-rev range.

Put simply, because it doesn't really work as well as it should and it takes two turbos. Let me explain...

Sequential turbocharging has been tried before. The JZA80 Supra (1993~1997; 2JZ-GTE engine) and the FD RX-7 (1992-1995; 2nd generation 13B turbo engine) are examples of cars which used two turbos in series. Neither car has the kind of torque curve we are shooting for.

There are actually four ways to arrange two turbos although in practice only two of these ways have been used in automotive applications.

The first way is putting two turbos in parallel. You get the flow capacity of a larger turbo in two smaller ones with lower inertial. Reducing inertial reduces lag much more than not splitting the exhaust energy from all cylinders between two turbines. This is also a convenient arrangement in a Vee type engine since you don't have to route the exhaust from both banks to one turbo. This is by far the most common way to arrange two turbos.

The second is to put all the exhaust through one smaller turbo first, then feeding the exhaust of the first turbo (both the turbine exhaust and the wastegate exhaust to a larger turbo. The output of both are combined in parallel and fed to the intercooler. This is the arrangement we encounter whenever a car is said to be sequentially twin turbocharged. The advantage is that the smaller turbo provides boost with reduced lag at lower engine speeds, while the larger one provides flow capacity to handle airflow demands at higher RPMs. The problem with this arrangement is that the exhaust routing is complicated and the results may not be as perfect as one may think. To begin with the first turbo and its wastegate restricts the airflow to the larger second turbo and also soaks away some of the heat. This degrades the performance of the second turbo. More importantly however is that while the arrangement allows boost to come on early, it really doesn't cure boost lag at mid to high rpms as well as we may expect because you stll need to spin up that larger turbo and it is hampered by the small one upstream to some degree. Also, if not handled well, the kicking in of the second turbo can cause a spike in boost and power midway through the powerband. In the end, the idea fell out of favor because the complexity and cost did not bring as much benefit as originally expected.

The third way is to parallel the exhaust feed to the turbos, but sequentialize the output. One turbo brings the boost pressure to say 15 psi while the other ingests 15 psi air and boost it up to 30 psi for instance. This allows both turbos to work within the efficient parts of their compressor map because a single centrifugal impeller is simply unable to reach about 30 psi without being rather inefficient regardless of its size. Jet engines use 7 to 15 compressor stages instead of one big one to reach a pressure ratio of 1:20~50 for this reason. In WWII it is popular to sequentialize centrifugal supercharger outputs in this manner. The Junkers-Jumo 213E engine in the high altitude Focke-Wulf Ta152H-1 interceptor is an example of a sequentially supercharged engine.

The fourth way is to combine both the second and third methods and sequentialize both the turbines and the compressors. I don't know think it has ever been done for automobiles although the concept is a staple in turbine engines with twin or triple spools.

Coming back to the LNF discussion. Basically, the point is that I wanted to keep it simple and stick to one turbine. Lag aside, this is also the most efficient way to get to 15 psi. If you look at the compressor maps, you can see that we won't gain much efficiency or map width going to 7 psi and sequentializing. Also, while a exhaust sequential arrangement would lead to faster boost onset you pay for it with upstream restrictions on the second turbo. There is also the issue that if we go with two turbos, they will have to be two very small ones. The GT25 is the smallest ball bearing unit available. On top of that, all the smaller units are not as efficient (by about 10%) compared to the GT25 and GT28s. Besides, all the suggestions -- except the turbo choice -- holds true whether you use one or two turbos. T

[thegriffon]

New twinturbo (true twin turbos, not one for each back of cylinders for eg) applications have been introduced for diesel engines, but have not been done in other automotive applications. They borrow an idea from active intake manifolds and have two different routes to the intake port at low speeds flow goes through the small turbine, at medium speed the large turbine is phased in, and at high speed the small turbo drops out. The GM-Fiat Twin turbo 1.9 works like this (9-3 1.9 TTid), as does BMW's 3.0 L twin turbo (535d etc). I believe a gas engine with a similar setup is on its way. VW instead favors what it calls a TwinCharger—combined mechanical and exhaust driven compressors.