Posted Thursday, February 2/06 in Mods & Tests

Anyone with a passing interest in engine performance is probably familiar with the concept of the cold air intake (CAI), where the goal is to feed the coolest possible air into the engine. Since cooler air is denser (contains more oxygen by volume), a modern engine will compensate by injecting more fuel into the mix to retain a proper air/fuel mixture. The result is more power at a given throttle opening (relative to warmer air).

The idea behind a warm air intake (WAI) is based on the same underlying principles, but its goal is 180 degrees in the other direction: heating the intake air and decreasing its density to reduce power and boost the engine's overall efficiency.

A WAI set-up is easy to make, so about a month ago, I put one together for my car. It immediately suceeded in increasing my intake air temperature significantly, and I ran a controlled-as-possible test to see what it did for MPG...

---

WAI test overview:

• the theory behind WAI
  
• making the WAI
  
• weather conditions & test route
  
• test results: CAI vs. WAI
  
• observations & conclusions

---

The theory behind WAI

Effect of air intake temperature (IAT) on fuel trim: less than 1 is leaner than stoichiometric. Shown for a Honda B16A2 engine (96-00 Civic Si) - Thanks CRXmpg.com

• Some cars' engine control units (ECUs) adjust the air/fuel mixture based partly on readings from the intake air temperature (IAT) sensor. The warmer the air, the leaner the mixture, up to a point:

  "OBD2 systems will attempt to run the system fairly lean ... Approximately 70 F is generally used for 0 correction, temperatures above will result in less fuel to compensate for the lower air density, and conversely temperatures below 70 F will result in steadily richer (by volume) mixtures." - (source)

• Some cars (notably certain Hondas) have a "lean burn" mode, where the engines are designed to run on significantly leaner mixtures under certain conditions than regular engines. A prerequisite for entering lean burn mode on some of these engines is reaching a minimum IAT. The computer needs to "see" a minimum temperature, or lean burn mode is not used.
• Reducing the density of the intake air by heating it effectively reduces the power of the engine at a given throttle opening (relative to the same engine with a CAI). To do the same work as the CAI-equipped engine, the throttle of the WAI engine must be opened wider at a given RPM. This increases engine efficiency by reducing throttling or pumping losses:

  "The air is less dense, so you get less horsepower at the same throttle opening, thus, you have to open the throttle wider to let in more air and get the horsepower that you need. That increases the efficiency because one of the primary causes of the well-known part load inefficiency of gasoline engines is the throttle loss." - (source)

• WAI possibly contributes to marginally better fuel vaporization in cold weather.
• WAI may contribute to quicker engine warm-up, reducing (to a small degree) fuel consumption between a cold start and normal operating temperature.

Some of these situations only apply when the ambient temperature is very low, and others depend on how the car's computer is programmed to respond to the signal from the IAT.

However, the benefit of reduced pumping losses should increase efficiency regardless of ambient temperatures or computer response.

Removing the snorkel between the filter housing and the stock cold air intake (CAI)

(It is even possible to control engine speed through varying the temperature/density of the intake air as a primary control, when varied to a great degree. Students at the University of Southern California built a throttleless engine on this principle using variable intake air temperature and leaning of the intake charge to control power output. - source).

Making the WAI

This was the easy part, and I happened to have all the bits and pieces in the garage already.

I started by removing the flexible snorkle connecting the air filter housing to the stock CAI tube (which leads through the side of the engine compartment and terminates in a resonator box inside the right fender).

I attached a length of heavy plastic tubing to the filter housing. For this, I used a length of shop vac hose (the ends of which were worn out, so I had bought a new one). By chance, its inside diameter was a perfect match for the end of the filter housing.

I routed the hose around to the front of the engine to the exhaust manifold and slipped a length of steel exhaust pipe in the end of the hose. The pipe is the part that actually rests on the manifold. But won't the plastic hose melt? It's far enough from the manifold that it has not melted yet, anyway - in 1500+ km of driving. I cut the end of the steel pipe on an angle to create a large opening for hot air to be drawn up from between the exhaust manifold and engine.

The pipe is held in place with heavy wire (coat hanger) fished under the manifold and twisted up around it from below.

- Total WAI cost: $0.

As can be seen in the photos, I eventually added an aluminum foil shroud over the works to trap as much hot air as possible around the opening. At first I didn't use the foil, and I saw temperatures like these:

- 30 F / -1.1 C ambient
- 75 F / 23.9 C max intake air temp (IAT)

After adding the foil shield, the IAT measured on 2 different days were:

- 25 F / -3.9 C ambient
- 105 F / 40.6 C max IAT, idling in traffic
- 95 F / 35.0 C IAT @ 80 km/h

- 45 F / 7.2 C ambient
- 112 F / 44.4 C max IAT, city
- 104 F / 40.0 C IAT @ 80 km/h

The engine/coolant temperature remained close to its normal range, with max values occasionally 5 F / 2.8 C higher than when using the CAI.

Caveat experimentor: overheating the intake tract may cause component damage, as pointed out by a member of the MaxMPG Yahoo forum.

Weather conditions, test route, methodology

• weather conditions were as follows, based on info from an online weather station located a coupld of km from the "test route":

  time....wind-dir...wind...gust...temp-F

  12pm......ssw.......10.....10......39
  01pm......ssw........8......8......38.1
  02pm......ssw........7......7......39.6

  (wind data in knots; the test went from approx. noon to 2pm.)

• "test" road: a nearly level, nearly straight 6.6 km / 4.1 mi. stretch (13.2 km round trip) of highway
• bi-directional run averages are compared (to cancel effect of wind/grade)
• speed was 89 km/h - the cruise control was set once, and cancelled with the brake pedal between runs; the "resume" feature used for subsequent runs (cruise control consistency was verfied by noting "average speed" feature on ScanGauge after several runs). The car was up to steady speed on cruise control when passing a fixed "start" or "stop" marker; ScanGauge "avg" logging was either reset (at "start") or recorded (at "stop" marker)
• an A-B-A-B test order was used: 3 CAI runs; then 3 WAI runs; then 2 CAI; then 1 WAI
• when switching between A and B, the car was driven for 10 minutes to allow the intake tract to either cool down or warm up to a steady temperature before returning to the "test course"
• fuel used during testing was 6.9 L, so the car became lighter by that amount from run 1 through run 9 (includes the cool down/warm up drives)
• operating temperature: the car was warmed up prior to the first run with a block heater (1.5 hrs), and 30 minutes of mixed city/hwy driving

Test results: CAI vs. WAI

@ 89km/h / 55.3mph -- East/West -- units: km/gal(US) -- intake temp(F)
- - -
CAI - run #1 --- 82.9 E ----- 74.1 W ----- 78.50 avg --- 54.0 F
CAI - run #2 --- 83.3 E ----- 74.8 W ----- 79.05 avg --- 53.5 F
CAI - run #3 --- 83.6 E ----- 75.2 W ----- 79.40 avg --- 55.7 F
CAI - run #7 --- 83.0 E ----- 75.8 W ----- 79.40 avg --- 55.5 F
CAI - run #8 --- 82.5 E ----- 72.6 W ----- 79.25 avg --- 54.3 F
- - -
WAI - run #4 --- 82.2 E ----- 75.6 W ----- 78.90 avg --- 114.3 F
WAI - run #5 --- 84.9 E ----- 75.5 W ----- 80.20 avg --- 117.3 F
WAI - run #6 --- 81.1 E ----- 75.6 W ----- 78.35 avg --- 116.0 F
WAI - run #9 --- 86.2 E ----- 73.4 W ----- 79.80 avg --- 115.0 F
- - -
Average of CAI bi-directional runs = 79.12
Average of WAI bi-directional runs = 79.31

Observations & conclusions

The grain of salt: the margin of error for the CAI group is +/-0.17 km/gal; for the WAI group it's +/-0.42 km/gal.

So there you have it. For this car, under these specific test conditions, the difference between the WAI and CAI averages is so small as to be statistically insignificant. (And even if it was statistically significant, practically speaking there's no difference between 79.12 and 79.31 km/gal anyway.)

So... why didn't it work?

It's always disappointing to put this much work into a test and not see good results. Even more so when others swear by the effectiveness of the modification:

In particular, the folks over at the Honda Insight forum - InsightCentral.net - insist this hack works for them. They say that without a WAI, their winter MPG drops through the floor relative to warm weather performance. (For examples of their forum discussions on this mod: thread 1, thread 2, thread 3)

Of course, the difference between the Firefly/Metro and the Insight in this respect is the Honda's lean-burn design. Presumably the Insight's IAT sensor needs to see a minimum temperature before lean burn happens.

And, what about the idea that warmer air should result in leaner fuel trim? All OBD2 cars do this to some degree, though it's not as pronounced as a "lean-burn" system like in the Insight. I looked into this after the fact and learned that the IAT sensor on the Metro 1.0 engine apparently doesn't even participate in the fuel metering system. According to the car's factory service manual, the only players in the fuel metering game are:

• the manifold absolute pressure (MAP) sensor
  
• the engine coolant temperature (ECT) sensor (aka the water temp sensor WTS)
  
• the throttle position (TP) sensor
  
• the heated oxygen sensor (HO2S)
  
• engine speed (RPM)

But even ignoring fuel trim effects, the reduced pumping/throttling losses in moving less dense air through the engine should have helped fuel consumption to some degree. Why doesn't it show up in the results? Perhaps because:

• The throttle is open slightly wider, and the effective load is higher. The ECU may be compensating by enriching the mixture
• The reduction in pumping loss is insignificant to overall efficiency.

Other potential factors masking the benefit of less dense intake air could be:

• The effect of a WAI may be most apparent at very low ambient temperatures (ie. colder than the fairly mild day the test was run)
• The test measured effects at a very narrow throttle range (fixed speed); at wider, narrower or variable ranges the effect may be more pronounced (unfortunately I can't control a variable throttle test)
• It was mentioned over at teamswift.net that the IAT sensor may have a role in controlling ignition timing, and if so, the efficiency gains may have been offset by less ignition advance. I haven't been able to verify that yet).

But... I'm going to leave the WAI on anyway, at least until early summer. I've decided not to exceed 130F on the AIT sensor.

Why? Because, after all, it didn't HURT my mileage. And it may help the engine reach normal operating temperature slightly faster from a cold or warm start, which would still aid efficiency to a small degree.

The biggest mystery that needs to be solved is the exact role of the IAT sensor in this car. I may find that I have to "tune" (spoof) the sensor if it's involved in counteracting the WAI gains.

Of course, if the results of this experiment had shown an efficiency gain of the WAI, I wouldn't particularly care to know all these details, but since it didn't work in the face of multiple theories suggesting it should have... I want to know why.

So - to be continued...

Resources:

• Throttleless Premixed-Charge Engines - University of Southern California

• Intake Air Temperature Tweaks - odb2crazy.com

• Peter O's Warm Air Theory - Swedish Bricks (Volvo) discussion list archive

• WAI thread 1, thread 2, thread 3 - InsightCentral.net discussion forum