[Closed] Newer car engines? Ecoboost or puretech anygood?

I believe the current crop of small turbo petrol engines have a surprising amount of real-world power but the fuel consumption suffers badly when you inevitably work them hard. Explanation here:

http://www.telegraph.co.uk/cars/news/age-engine-downsizing-says-volkswagen/

The little Fiesta Ecoboost goes really well... But... it's still a Fiesta, so still a little cheap and road-noisy.

I believe the current crop of small turbo petrol engines have a surprising amount of real-world power but the fuel consumption suffers badly when you inevitably work them hard. Explanation here:
> http://www.telegraph.co.uk/cars/news/age-engine-downsizing-says-volkswagen/

I wouldn't rely on a Volkswagen press release as an authoritative source on emissions 😀

I have a big car with a 1l turbo petrol engine. I drive it up [url= http://www.cyclingweekly.co.uk/news/racing/tour-de-france/cote-de-ripponden-bank-10-greatest-tour-cycling-climbs-yorkshire-128157 ]Ripponden Bank[/url] every morning, and it's never struggled to keep up with anything in front of it.

I believe the current crop of small turbo petrol engines have a surprising amount of real-world power but the fuel consumption suffers badly when you inevitably work them hard.

I recently hired a 1l Focus ecoboost. Despite not driving it like a rental, I was not that impressed with the reported MPG.

I've had a 1.0 ecoboost Focus for a year now (6 speed 125bhp). Absolutely brilliant engine. It'll do 55-60 on the motorway fully loaded with a bike on the roof, more if you're really drivin' Miss Daisy. In horrible slow crawl city traffic it'll get 30-35.

Driving it like I stole it across country B roads I can't get it to go below 30mpg really. The engine is very clever about when it's on and off boost. I would buy another, and would love the 140ps Fiesta version as a second car if they did a 5 door version (and maybe more sedate looking).

[u]Without reading the link below[/u], when you're flat out most engines are pretty efficient, so you won't make much difference to that state.

LOL. It says that they get hot, so the computer makes them run richer to cool down which sends fuel consumption and emissions where you don't want them. An emission admission from VAG.

I've driven 2 Ecoboost Car, a Fiesta and a Focus, they were pretty good actually.

I think the jury is still out on the long-term ownership deal, some people point to all the bits and bobs they need to make a 1ltr go like a 1.6 or even more and compare them to modern Turbo Diesels, lots to go wrong etc, but really that sort of thing is for the 4th or 5th owner at 10 years and 100k miles to worry about.

Oddly, I also know that whilst they're more efficient than their forbearers, they're much heavier - forced induction meant an iron block rather than Alu, added to the turbo and pipework needed means a heavier engine - but that was droned to me by a guy into kit cars, not really all that important to most people, but a pain shared and all that.

P-Jay

Oddly, I also know that whilst they're more efficient than their forbearers, they're much heavier - forced induction meant an iron block rather than Alu, added to the turbo and pipework needed means a heavier engine - but that was droned to me by a guy into kit cars, not really all that important to most people, but a pain shared and all that.

That's not correct though. The 1.0 ecoboost engine weighs 75kg iirc, about the same as 1.0 4cyl alloy engine. The 1.6 which it's comparable to is something like 120kg. It's also very small so it sits further back in the car improving the steering.

[img] [/img]

My daughters Fiesta has the 125bhp version of the 1 litre Ecoboost. It's very impressive to drive - reminds me of the hot hatchbacks I had in the 90's in fact. It's capable of pretty impressive fuel consumption but because it's reasonably quick it tends to get booted somewhat, although still returns 40+mpg even like that.

I have a Fiesta 1.0 125... good little engine, performs way beyond what you'd expect of a 1.0, but if you have a heavy right foot you won't see even 40mpg. It's fine in the Fiesta, but doesn't have enough oomph to pull the bigger Fords around though - it doesn't cope in a Focus.

The 140 version is apparently the most powerful 1.0 ever fitted to a stock production car, but there weren't many on the 2nd had market when I got it last year.

LOL. It says that they get hot, so the computer makes them run richer to cool down which sends fuel consumption and emissions where you don't want them. An emission admission from VAG.

That bit didn't quite ring true (and it's not a VAG quote, it's from Reuters). The direct injection engines work on the principle that you can have part of the combustion chamber with a stochiometric mix and part just air, so as the mix burns it pushes down on the air, which pushes on the piston. Which is more efficient than running the engine with less air like a conventional engine. It also avoids NOx because the excess air (where NOx comes from) isn't heated. When you floor it, it reverts back to normal and just fills the combustion chamber with a stochiometric mix (i.e. it's richer, but not necessarily over fueled?

Oddly, I also know that whilst they're more efficient than their forbearers, they're much heavier - forced induction meant an iron block rather than Alu, added to the turbo and pipework needed means a heavier engine - but that was droned to me by a guy into kit cars, not really all that important to most people, but a pain shared and all that.

Remember that anything related to kit cars is basically dealing with 10+ year old engine technology though. And the boost needed to get 60% more air into an engine (i.e. 1.6 performance form a 1.0) isn't anywhere near as extreme as what kit car builders would do.

previous car was a 1.2 old school and I ended up having a queue of traffic behind me.

Ignore the engine size. 110bhp is 110bhp, regardless. And it's plenty to drive around in. Old 1.2s had like 55bhp.

I think that the loss of efficiency under different driving conditions comes from the fact that when it opens up the turbo vanes to give you more power it impedes the exhuast flow. So accelerate slowly, you don't use the turbo and it's efficient.

I think, anyway.

But remember the other thing about having a turbo is that you can control when in the rev range it gets extra boost. My dad's Golf is only 85bhp yet it has plenty of useful torque - much more than a NA 1.4 engine.

mike_p

It's fine in the Fiesta, but doesn't have enough oomph to pull the bigger Fords around though - it doesn't cope in a Focus.

It's absolutely fine in a Focus. As Molgrips says, 125bhp is 125bhp. And 150lbs ft is still 150lbs ft whether it's naturally aspirated or forced induction. More is always better but I'm coming from a 300bhp Subaru to the 1.0 Focus and to be honest there aren't many times I miss the Subaru.

Well having driven both I don't agree, so there 🙄

We've got a 1.0 ecoboost Fiesta. The little 3 cylinder engine is a bizarre beast, it shakes around all over the place if you run it with the bonnet open. It's light enough to zoom around town fairly comfortably and when you rev it enough to get the little turbo to kick in, it has a decent zoom.

We had a Mini Cooper S previously, with 180ish bhp. It had tons of zoom when you got going but I curiously found it a real drag at slow speeds, hill starts and things were a pain in the arse. 180bhp is 180 peak bhp, it doesn't mean it's great throughout the range.

Same as jimjam, we have a 1.0l Ecoboost Focus (the 125bhp version). Very impressed with the performance. The engine was designed to replace the Zetec 1.6 petrol but it drives more like a 1.8 turbodiesel (slightly more responsive though).

If you do lots of motorway driving and keep it at a sensible speed (70mph) you can get 45mpg easily. I've had it over 50 a few times on longer trips.

Around town and on country roads I get about 40mpg. The Mrs gets 35 but she's heavy footed.

We've noticed that the engine prefers higher octane fuel (Shell V-Power / Tesco Momentum) and pulls noticeably better using these types of fuels.

If you want better mpg then get a diesel.

The 140 version is apparently the most powerful 1.0 ever fitted to a stock production car, but there weren't many on the 2nd had market when I got it last year.

Apparently the 140 version is exactly the same engine as the 125, just remapped. You can buy aftermarket tuning for the 125 for a couple of hundred quid.

Oh that's one thing to note about the smaller Fiesta as well. Ours merrily sits at just over 50 mpg at about 50-60, if you head up to 70 this tails back off to mid 40s or so. Still impressive, it just feels like it's not really designed for lots of long motorway journeys.

As for the weight thing, I truly don't care but...

1.6 Sigma Duratec ti-vct is 90kgs dry

1.0 EcoBoost is 97kgs dry

Wether than qualifies as "much" heavier I don't know, but it does weigh more it seems.

I guess the EcoBoost and it's various power outputs was the direct replacement for the smaller engines in the previous line too which would have been lighter than the 1.6 and it's variable valve timing kit.

Jus sayin'

1.2vti Peugeot 208. Good cheap commuter. Pretty much 50mpg with a miserly 89bhp. Mostly country roads and a bit of round town.

Bit rattly but not as bad as some small cars.

Got a 2008 as a courtesy car last service and it had the same 1.2 engine, car didn't feel much bigger but it was only getting around 44 mpg.

The 40ish mpg posts are interesting.

Mrs_oab has Seat Ibiza ST 1.4 16v - the old engine. Yes it lacks oomph (95ish bhp I think) - but does a long way over 40pmg on a run and 35ish all day long.

It's a car that was much cheaper to buy than 1.2tsi, has no turbo, fancy pants sensors and more.

I think the financial or environmental 'gains' of newer engines don't stack up, especially when extra purchase and maintainance costs are factored in.

We have the 1.6 150 version in a Tourneo Connect. It certainly nippy but not that economical.

So now I'm looking at the Superchips remap....£399 gets 151bhp and 184lbs ft. No need. No need. Pointless.

What are the extra service costs on an ecoboost engine?

Long term - all the sensors, dual mass clutch etc. much like the current crop of diesels.
Day to day servicing, no difference, more ongoing maintenance/wear out costs.

Most petrols have those don't they?

Currently got Cactus (it's a lease car and was cheap! ), It's the 1.2 turbo, 3 cylinder 110hp one. On my run to work does around 53mpg, Infact it more or less stays at that. My previous car was a 1.6 diesel fiesta and it done 54 mpg on the same run. I do drive like Mrs Daisy though, as generally never in a rush to get to work.

It's nippy enough and for general kicking about in its fine, it's seems higher geared than the fiesta was and does a better job on longer motorway journeys too.

I will confess though, it's just a tool to do a job in my eyes, so maybe my view is not the most informative view to take!

Agreed, but our engine is old enough not to have them.

The point being I wonder if all the new petrol 'gains' are not all they cracked up to be - just like the diesel gains, especially with extra costs later in car life.

The costs arne't as bad as you think. Cars failing now are 10-15 years old. You won't know how much better new cars are until 10-15 years time.

Mrs Inbred has a 110bhp 1.2 fabia dsg. It never struggles to keep up with any car. In fact it's so good I thought about getting a fabia estate with the 1.2 110bhp engine.

What are the extra service costs on an ecoboost engine?

Newish engine so no idea how they will behave in 5-10 years time.

Apparently the turbocharger on the ecoboost spins at some ungodly RPM so that could possibly be the first thing to go on them.

Cars are a different kettle of fish nowadays. My grandfather always bought Ford and made them last 20+ years doing his own servicing. Can't do that nowadays with all these sensors and gadgets. Throwaway society - PCP finance, car for 3 years, then get a new one before the proverbial poop hits the fan seems to be the approach.

Apparently the turbocharger on the ecoboost spins at some ungodly RPM so that could possibly be the first thing to go on them.

The one on my dad's car is tiny. About 4" across.

My grandfather always bought Ford and made them last 20+ years doing his own servicing. Can't do that nowadays with all these sensors and gadgets. Throwaway society

That's total bollocks. You CAN absolutely do all that nowadays, you just need to buy a code reader. The car actually tells you what is wrong, you don't have to know a black art of carb adjustment or whatever. Just cos it's different and not the same as it used to be, does NOT mean it's rubbish!

And it's not throwaway either. There's no reason you can't fix any of it. Sure, a control module might fail, but I can go online and buy another one from a scrappy if I don't want to pay the dealer prices. It's really not hard at all.

I with Molgrips on this one.

For the most part new car's aren't as scary as people like to make them out to be. And if you really wanted to keep a 20 year old ford Mondeo on the road for some reason there's even plenty of aftermarket ECU's that'll do the job.

It's still mechanical things that'll kill most old cars. A set of tyres or a clutch still costs more than a lot of 15yr old's are worth.

TINAS; owner/maintainer of 10, 12 and 41 year old cars.

matt_outandabout

Agreed, but our engine is old enough not to have them.

Are you sure? My 2001 2.0 normally aspirated legacy had four o2 sensors in the exhaust and two water temperature sensors.

And it's not throwaway either. There's no reason you can't fix any of it.

My old Touran had the dreaded ABS pump failure thing. New unit, with sensor that had failed, was £1200, or refurb was £600ish with a three month warranty.

On my old old Passat when similar happened, the mechanic reached under bonnet and swapped a sensor on the outside of the abs pump. Cost iirc was £30.

My Galaxy needed a new front bearing - only available pre installed into a hub. So a £20 bearing actually costs £150+

I've got an Octavia 1.4Tsi. Around 125bhp I think. Pulls better from 1500rpm than my old 1.8 non turbo petrol Mondeo though doesn't go quite as quick at higher revs. Prefer the better low down pulling power of the 1.4Tsi for everyday driving.

My Mondeo used to get 32-38mpg. The Octavia gGets 42-52mpg. A 2012 model, 46k miles on it. No problems so far. At 10k miles per year I think I save about £500 a year in fuel compared to the Mondeo. Road tax was about £80 a year less as well. Though any road tax advanges the small turbo petrols have will vanish in April when it goes to a standard £140 for new registrations.

We are replacing Mrs IRC's car at the end of the year and the B-Max 1.0, 125bhp is the favourite at the moment.

Matt OAB - you're not really talking about the same thing. I could give you a list of simple failures that are an easy fix on a new car just as an old one. I could also remind you of the days when 100k miles was a lot on a car, and you'd have to watch for big ends and piston rings failing, people with tin foil in their radiators and having to carry a can of WD40 in case it rained too much.

I've got an intermittent fault on my ABS control module. I know this because the car told me it was there. In the old days I'd never have known until it was possibly too late and it failed in an emergency. If I even had ABS that is. I am looking at refurb pumps on ebay for £90, and I'm told you can simply replace the control module without taking the pump off.

Misplaced nostalgia really ticks me off.

My Galaxy needed a new front bearing - only available pre installed into a hub. So a £20 bearing actually costs £150+

£20 part, but you need an expenisve set of presses and a mechanic to replace it.

Had a 1.0 Ecoboost Fiesta for 2.5 years - due for replacement in August - a lease car - a car isn't a thing to keep in my mind, it is a tool to do a job so ownership isn't high on my priority list.

When I ordered the Fiesta it was ideal - bit bigger than previous car, bit more oomph and fuel economy was sounding great.

Real world saw me getting about 52mpg for the first year (with roof bars and 2 racks on the roof permanently); it has had 2 services since then and I appear to be getting about 43mpg now (same roof bars but with 3 racks on the roof) - I think I'm driving it the same way, but I might be being a bit heavier footed and not really realising.

However, it feel good performance with 2 adults and a child in it, but when you load it up (3 bikes, long weekend camping), you can really feel it working - it isn't as quick (noticeably not as quick).

I sit with the speedo reading at 70mph on the clock, which is about 67mpg according to my 2 GPS devices) - over the last 6 months, the instant MPG is reading a lot of 20s and low 30s; previously it would be low 40s and some 50s - this isn't my driving as the cruise control is switched on and the same stretch of road is being driven - so I suspect the engine and whatever parts aren't quite as efficient as they were.

It is sitting on 30k, so I think average mileage and it gets between 15 and 40 miles a day on it.

It has been a great car, but I'm not needing something bigger - I would really like a Focus Estate - with the 125 Ecoboost engine - my concern there though is will I notice the lack of oomph even more as the car is slightly heavier (doesn't appear to be massively heavier - just over 1 ton to about 1.2 ton (so 2x6 foot rugby players in the car - actually probably is noticeably heavier!)), it will have the 3 racks on the roof but there should be a bit more space for the camping/holiday/weekend gear where it isn't crammed in.

So how much of a difference will I notice? I'm not a boy racer; I rarely drive in a spirited fashion, but I do prefer driving at the higher end of safely-permitted-to-do-so.

Needless to say, I'm impressed with the wee Ecoboost engine, it has been very good, however, it doesn't seem to be as efficient as it was. I'm putting this down to a slight change in my driving behaviour but also as the engine is ageing and things aren't quite as tight and new as before. (says me who is not mechanically minded when it comes to cars at all!)

however, it doesn't seem to be as efficient as it was. I'm putting this down to a slight change in my driving behaviour but also as the engine is ageing and things aren't quite as tight and new as before

It's not old in the least. It should be just loosened up by this point so you should be getting the best MPG.

If my MPGs had dropped half that much I'd be down the garage immediately. It's a huge drop.

says me who is not mechanically minded when it comes to cars at all!

When was the last time you checked tyre pressures? Check them cold.

What no one has said so far is the benefit of the turbo.

On 1.0's or 2.0's they make petrols more driveable ie give more low down torque, so for more day to day driving they just drive nicer than old petrols, which tended to need revving hard to get performance.

The Ford eco engine sounds pretty good when you rev it hard. Then again I've always thought 3 cyclinder engines do!

As to mpg, I can never get more than 44mpg out of our 1.0 100bhp Fiesta on a commute. However this compared to about high 30's in aToyota Aygo 1.0, high 40's in a 2.0 diesel or now low 50's in a 2.0 petrol hybrid.

My commute is 20 miles each way and hilly, so it kind of proves that small engines are not always most efficient

Checked every 2 months, they are sitting between the normal and economy pressure 36 instead of the 39 for most economic.

Got a 1.4turbo Astra (150bhp) on lease. Just about to hit 3000 miles. Average mpg since new is 40mpg with careful driving. If you press the accelerator down alot instant mpg hits single figures easily.
My old 2.4 200nhp desiel alfa would do 32mpg being thrashed all day.
Wasn't a service item on the 150bhp eco boost the engine block (100k miles iirc)?

It is 'old' as in 30k run in...I'd expect a slight improvement on mpg over the length of time I've owned the car as I think the driving behaviour should adapt to the car and engine...suspect I haven't adjusted as I expected and it is a slightly heavier foot than before.

It'd still be in the garage if it were me. You're talking about typical instant mpg whilst using crusie on the same road...

I've been looking at the fiesta, but round my way they are highly sought after by thieves for the st and the eco model, that hikes the insurance up, and as I have a ds3 at the moment was curiously looking at the new new c3 which looks pretty great, and have already found 4 nearly used ones just over £11k which is a bit of a bargain running the 1.2ltr 110bhp engine, just seems a step down from my 1.6 ds3 but saying that can't see any nearly new cars without these eco engines.

I like the engine, the turbo is good...however it needs a wee bit of revs to go comfortably.

I've got access to a couple of cars, all non-European and they all prefer changing gear sooner than the turbo...the efficient driving thing indicates a gear change and it is fine. On the turbo, 20mph in 3Rd isn't smooth in the turbo...when the turbo is kicking it is fantastic.

Well the VW one in my Dad's car is 4cyl rather than 3, which would make it smoother.

He gets about 53 or so on a long run, averaging about 48 with a mix of short town trips and longer runs.

[i]Got a 1.4turbo Astra (150bhp) on lease. Just about to hit 3000 miles. Average mpg since new is 40mpg with careful driving.[/I]

As a comparison my previous car, a Vectra 1.9 diesel AUTO averaged 43.5mpg (according to the trip computer zero'd on the day I bought it) over the 50k and 3 years I had it, with no attempt to drive economically.

I don't know about the Ford engine, but my OH has a 1.5 3cyl Mini Cooper, which does 42-45 mpg on her commute (computer says 50+). Mixture of roads, she drives as fast as the roads allow, I guess. It goes well and sounds good, too.

£20 part, but you need an expenisve set of presses and a mechanic to replace it.

Two trusted local garages and main stealer said they did not have the press, so always buy hub.

Having spent daft money on keeping the Touran and Passat going, I lean back towards the simpler Seat 1.4 16v and our 53 Yaris D4D for low cost parts and less to go wrong in the first place.

I take your point about cars in the past being knackered by lower miles over lifetime.

However, you are a good home mechanic, I'm not. New cars cost more at garages to fix.

That bit didn't quite ring true (and it's not a VAG quote, it's from Reuters). The direct injection engines work on the principle that you can have part of the combustion chamber with a stochiometric mix and part just air, so as the mix burns it pushes down on the air, which pushes on the piston. Which is more efficient than running the engine with less air like a conventional engine. It also avoids NOx because the excess air (where NOx comes from) isn't heated. When you floor it, it reverts back to normal and just fills the combustion chamber with a stochiometric mix (i.e. it's richer, but not necessarily over fueled?

That's not quite right - a petrol engine will always try and run at Lambda 1, ie the correct ratio of fuel to air so that all the fuel burns. If there were excess air it would be above Lambda 1 and temperatures and NOx would skyrocket. When the engine gets hot, especially with turbos, the Lambda is reduced (more fuel) with the excess fuel used to cool the exhaust system down. Typical Lambdas for this can go down to around 0.8, but high power vehicles will see it go down to 0.7ish.

That's not quite right - a petrol engine will always try and run at Lambda 1, ie the correct ratio of fuel to air so that all the fuel burns

No, it is right. It's a stratified charge engine. So under partial load the cylinder isn't a homogenous fuel/air mix, there's only fuel and air at one end.

https://en.wikipedia.org/wiki/Gasoline_direct_injection

New cars cost more at garages to fix.

They do often, but not always. But we are (usually) getting more for our money, it's not some kind of conspiracy.

It's still running Lambda 1 though, there is no excess air in the cylinder.

Lean burn, where there is excess air, is very very different to normal direct injection and requires extra exhaust after treatment to cope with the NOx from the significantly higher cylinder temperatures and excess air.

(Hate to do this, but my day job is calibrating petrol engines for major automotive OEMs so I can assure you this is correct)

Timely thread. Anyone have experiences of these things in larger vehicles?

My lease is up next month (actually, it was up on the Hyundai last year, I've got a stay of execution by way of a Passat CC till March). I'm looking at a Mondeo, and as I do far fewer motorway miles these days I was looking at the 1.5 Ecoboost rather than the traditional 2.0 diesel lump. Am I going to regret this?

I've got a 2 litre Ecoboost. It is splendid. 247 bhp and 254 ft lb of torque from 2 litres and averages 34 mpg with mixed driving. Pretty remarkable.

Back on-topic; I know people with the 1.0 litre Ecoboost, in Fiestas and they love them. Lots of MPG if you drive sensibly and plenty of grunt for such a small engine.

That Wikipedia page isnt quite describing direct injection correctly - it is basically saying DI engines are lean burn which isn't true. Essentially

Lean Burn EQUALS Stratified DI

Stratified DI DOES NOT ALWAYS EQUAL lean burn

It's still running Lambda 1 though, there is no excess air in the cylinder.

It's not the same lambda throughout the cylinder. So one end lambda is 1, at the other it's er.. inifinty.

Given your credentials, can you explain how it actually works then if I am wrong?

Stratified DI DOES NOT ALWAYS EQUAL lean burn

I think it says that, second half of first section - three modes of operation?

I'm seriously thinking about abandoning my "buy a car for < 4k" attitude and going new/nearly new. All this efficiency stuff doesn't seem great as cars age.

All my recent cars have had issues caused by EGR valves for instance. Not just the EGR valve itself, but clogged air intakes, swirl flaps and valves on direct injection engines clogging with soot and muck.

Current 2.0 TFSI has what sounds like a slight misfire as the valves are sticking and the only way to cure it is to take the top of the engine to bits and clean the valves by hand. 400 quid at an independent. 🙁

It's a tax on the poor!

It's not the same lambda throughout the cylinder. So one end lambda is 1, at the other it's er.. inifinty.

Given your credentials, can you explain how it actually works then if I am wrong?

You are right that it layers up the Lambda throughout the cylinder, but the reason for this is to give a rich pocket next to the spark plug. The total in-cylinder Lambda is still 1 for normal running in pretty much all petrol engines and so there is no excess air. DI allows you to time a rich pocket next to the spark plug for whenever is best for complete combustion of the entire cylinder charge. It can also then be used for other functions such as running a lean mixture throughout most of the cylinder and then a late main inejction to make a really rich pocket near the plug - this is used for knock control as the lean mixture cannot auto-ignite whereas the rich pocket is ignited by the spark plug. However, the in-cylinder Lambda is still <=1.

Lean burn is where Lambda > 1 and a few manufacturers have engines with it (Mercedes being the main one but I don't think it's in production any more). This runs with excess air in the cylinder allowing you to run with the throttle open and reduce pumping losses. This is great for efficiency but is horrific for NOx and as such you need to run with an LNT or SCR in the exahust to mop up the extra NOx created.

Component protection is then where Lambda < 1 for the purpose of cooling the exhaust using excess fuel. Lambda can drop to around 0.7 depending on the temperature drop required. In extreme examples, the DI can be used to inject fuel in the exhaust stroke to cool the exhaust. This allows the combiustion injection to be leaner (around Lambda 0.85) to give more torque, but there is still fuel delivered to cool the exhaust.

Hope that explains a bit - happy to answer anything else if required!

So it only saves fuel by allowing increased compression ratio and hence improving thermal efficiency?

Why don't they solve the NOx problem by filling the cylinder with EGR then injecting both air and fuel into a pocket?

I thought the idea was the fuel was injected very late (after TDC), so that the lean parts of the cylinder weren't hot enough to generate NOx?

(My background knowledge is in burner design, which is obviously a bit different to how a cylinder behaves, but still relies on adding fuel to the correct parts of the flame to minimise NOx)

So it only saves fuel by allowing increased compression ratio and hence improving thermal efficiency?

Essentially, yes. DI by itself in a normal, homegenous engine (i.e. 99% of those in the market) doesn't really do anything itself - what it does is enable other technologies which can help efficiency. One is higher compression ratio which DI enables by allowing greater scope in injection/ignition timing and also charge cooling through the evaporation of the fuel in the cylinder itself. Another is allowing use of larger amounts of valve overlap to capture EGR and allow a more open throttle (see below...). It also helps with more accurate metering, especially in transient situations where PFI metering gets fairly inaccurate

Why don't they solve the NOx problem by filling the cylinder with EGR then injecting both air and fuel into a pocket?

EGR is used for this, but for an efficiency benefit rather than NOx reduction. If the engine needs to operate at part load, then to run at Lambda 1 you need to shut the throttle partly to reduce the air load in to the cylinder - this causes pumping losses which reduce efficiency. By using EGR (mainly through valve overlap changes, but can also be an EGR valve like a diesel as well) you can 'pre-fill' the cylinder with EGR so that with a fully open throttle you still end up with a reduced air charge in the cylinder - open throttle = lower pumping losses = better fuel efficency.

In a lean burn engine, the whole point is that you run with the cylinder with excess air to allow leaner conditions with the fuelling required for combustion so this reduces your scope for EGR use, however it is still used. The in-cylinder temps in a gasoline engine are significantly higher than a diesel though so it's much harder to use EGR to lower combustion temperature to reduce NOx, hence the need for aftertreatment.

I thought the idea was the fuel was injected very late (after TDC), so that the lean parts of the cylinder weren't hot enough to generate NOx?

In a gasoline engine you wouldn't be able to put the main injection that late and make any torque - the air and fuel needs to mix before it can be ignited around TDC (obviously depending on spark timing). Even with a stratified charge system there is still a good amount of charge mixing, you just keep injecting fuel later and later to put a richer pocket near the top of the cylinder.

In a normal homogenous charge engine, because all excess air is used up in combustion (or should be!) the NOx contribution is tiny over a cycle so there is no real need for directly calibrating for it (as such, clearly we make sure it is as low as possible, but CO and HC emissions are much more significant).