Posted Monday, November 28/05 in Mods & Tests

Unencumbered by the thought process, I bought an $80 CDN "performance" air filter shortly after I got Firefly #1 (and that filter recently migrated to the new Blackfly).

My (minimal) reasoning at the time was that anything that reduces the amount of energy needed to pump in intake air should be beneficial to fuel economy. And that's what performance filters are supposed to do - reduce restrictive airflow into the motor compared to stock paper filters.

Since then, I've learned a bit more about engine air flow which cast some doubt on my original "thinking". So I was back to wondering, will a high-flow filter really result in higher MPG? To find out, I put 3 filter types to the test using the recently acquired ScanGauge.

---

Filter test overview:

• 1st of 2 parts (go to part 2)
  
• weather conditions & test route
  
• filter types
  
• test results at 80 km/h (50 mph)
  
• possible explanations & plans for part 2

---

The weather this past weekend was finally right for running the experiment. I had selected a route weeks ago (a nearby long, straight & level stretch of 2-lane highway), but the weather this month has been really blustery, blowing 10-20 knots (18.5 - 37 km/h / 11.5 - 23 mph) practically every day. Gales of November to be sure.

A steady wind would be no problem, since I was going to be comparing data that was averaged over 2-way runs. (Averaging will cancel out the effect of a steady breeze - and changes in grade). But when is the wind speed ever constant? Certainly not this November. A couple of poorly timed, strong gusts could deliver enough energy to skew the results.

But the winds dropped to a relatively steady 4 or 5 knots on Saturday, so I headed out.

Weather conditions & route

Conditions were light breeze (SW < 5 kts); 28 F / -2 C steady ambient temperature; "damp" road surface.

Runs were 2-way, easterly and westerly, (8.6 km / 5.3 mi each way) and averaged to cancel effects of wind/grade. The cruise control was set once, at 80 km/h / 49.7 mph, and cancelled with the brake pedal after each run, so the same speed could be "resumed" for each filter type. The car was up to steady speed before passing fixed start & stop markers, where the average fuel economy function was reset or recorded.

The test: three filter variations

Using the ScanGauge, my main goal was to determine whether there were any differences in reported fuel consumption at a fixed, cruise controlled speed using 3 different filter types:

• OEM paper (with < 2500 km on it);
  
• K&N "high flow" (< 7000 km on it);
  
• and "none" (for kicks)

The test mule was my '98 Pontiac Firefly (Geo Metro), 1.0L 3-cyl, 5-speed, which had been warmed up beforehand with a 35 minute city & highway drive.

I set the gauge to read "kilometers per US gallon" because that combination gives the highest data resolution (smaller changes will show up better compared to miles per gallon).

To remove potential effects of the restrictive nature of the stock intake plumbing upstream which could mask filter differences, I disconnected the snorkel between the filter housing (which is right on top of the throttle body) and the tube leading from the air box/silencer in the fender. That way we're testing just the filters here. As a result, the engine was breathing 33 F / .6 C air (as reported by the gauge) drawn through the filter from the engine compartment (as opposed to normally from inside the fender).

Runs & results

Since I already had the K&N in the car, I started with 2 bi-directional runs to see what kind of margin of error there might be. Then I tried the other filter combinations, for a total of 8 one-way or 4 bi-directional runs.

@ 80 km/h / 49.7 mph --- East/West --- units: km/gal (US)

K&N filter - run #1 ---- 80.4 E ------ 75.5 W ------ 77.95 avg
K&N filter - run #2 ---- 79.9 E ------ 75.9 W ------ 77.90 avg
OEM paper -------------- 80.2 E ------ 75.4 W ------ 77.80 avg
No filter -------------- 80.6 E ------ 76.3 W ------ 78.45 avg

From this small sample, it looks like there's practically no difference between OEM and K&N (the difference between the two is barely greater than the margin of error between the 2 K&N runs).

Possible explanations; plans for Part 2

Remember, this is a tiny sample, so generalizing is risky. Repeat - risky. Every observation made from these results should be taken with a grain - no several grains - of salt.

With salt shaker firmly in hand, it appears there may be some restriction in air flow above the throttle plate/butterfly valve - as shown in the biggest change in average readings on the runs when no filter was present.

This would saltily suggest that a less restrictive filter should help reduce pumping losses and improve efficiency.

But since I'm not planning to drive without a filter, we're left looking at the practically meaningless difference of 0.125 km/gal (.08 mpg US) between paper and K&N (the average of the 2 runs).

Regardless of the sample size, it's important to note that this test says nothing about the relative performance of filter types at wider throttle openings. In fact, I'd go out on a limb and say that at wide open throttle (WOT), the K&N filter would probably deliver more air (and therefore better efficiency due to decreased pumping losses) than OEM paper.

But I don't drive at WOT, and I'm not really concerned with how much fuel I can save at 150 km/h. No, this test was designed to find out if there's any difference at fractional throttle openings at lower speeds, where I spend the majority of my time.

So did I waste my money on the filter? I need to reserve judgment until I get a chance to go out and try this again. I ran out of time on the weekend (I had actually started to repeat the test at a higher controlled speed, but a number of factors got in the way of reliable results). So I'll have to post my answer in part 2.