Measuring power on the dyno is a no-brainer: strap the car down, bring it to temperature, and floor it in third gear three times over and average the results. The process becomes slightly more complicated with automatic transmission-equipped cars, as the pull has to start higher in the rpm band to prevent automatic downshifting. Except for a spike when the 166,000-mile-old torque converter slipped, my Integra subject car achieved an average of with a relatively flat, an average air/fuel curve of 12:1 at wide-open throttle. Because the dyno never moves, offers constant altitude, resistance, weather and straight-line testing (i.e. the rollers), as well as incurring no wind resistance or traffic, testing on one eliminates many of the variables that could sway results of measuring fuel consumption on the street. By working to eliminate what variables do exist and recording their differences, testing and tuning on a dyno can give very accurate, reliable and reproducible results. The downside is that the results achieved on the dyno will never reflect a real-world value for fuel consumption-we'll need to road test for that. Simply, if all tests are conducted under the same circumstances, an absolute "working value" for consumption can be determined, and its differences reliably measured from test to test.
To determine a working value for fuel consumption, we first had to establish our "control set," to ensure all our variables would remain consistent throughout testing. To measure fuel consumption, we decided that drawing from an external fuel pump and tank would be most accurate, since it could be easily weighed, before and after a dyno run, to see how much fuel was spent. Our makeshift fuel tank setup included: an identical fuel pump as the one in the car (pulled from a donor Integra LS), suspended by a cut section of plastic container and placed inside an enclosed plastic container that would be the fuel tank. Since we wanted the test to reflect how the car was receiving and spending fuel in its driven condition, all necessary components of the car's fuel system had to remain intact for testing that were used in daily driving. Fuel delivery and return lines were made to attach to the car's fuel filter, and lengthened enough to allow the makeshift assembly to be sat on a scale for measurement. Since fuel flow in a Honda is controlled at the fuel rail by a mechanical pressure regulator, the pump is either "on" or "off." Because of this, we were able to hard-wire the pump to a 12-volt power source, and let the regulator decide how much fuel the injectors needed-what fuel wasn't needed was simply returned to our tank assembly via the return line.
At rest, with slightly less than a gallon of 87-octane gasoline, our assembly weighed just under seven pounds. This was the weight of not only the fuel, but the pump and tank assembly, lines and wiring as well; since none of this will change throughout testing, we saw no need to weigh it alone. With the assembly on the scale, and the car on the dyno, it was time to start the car, and allow it to reach temperature by idling for several minutes. Once it had done so, the test was about to begin. Wanting the test to reflect highway driving conditions as normal as we could reproduce them on the dyno, we decided that testing would be conducted with the car in 3rd gear, at a wheel speed of what registered as 60 mph according to the odometer and 59 mph to the dyno, and with an engine speed of 4000 rpm according to the tachometer and 3700 rpm on the dyno. After a few minutes of holding the throttle steady to achieve a constant speed, the timer was tripped, and at the same time, the weight of the fuel tank assembly recorded to determine a starting point. After holding the throttle as steady as possible for 15 minutes, the engine was killed, and a new weight of the fuel tank assembly was recorded. The process was repeated three times over and averages of all data were made to determine our final working numbers. Our results between runs was surprisingly consistent; the first run saw 32.9 ounces of fuel spent over 14.8 miles traveled in 15 minutes, while the second and third runs saw 33.1and 33.0 ounces of fuel spent over 14.9 and 14.8 miles, respectively, each in 15 minutes. We attribute the differences in results between each run to human error with throttle modulation-it's harder to keep a constant throttle for 15 minutes at a time than one might think!
Doing a little research on the weight of gasoline, we find that gallon of 87-octane gas at our testing temperature of 72 degrees Fahrenheit weighs about 6.25 pounds. Using our working average of 33.0 ounces (or 2.0625 pounds), we learn that we've spent .33 gallons of gas over the average 14.833 miles traveled on the dyno, resulting in a working average consumption of 44.95 mpg. While it seems high, this number hasn't been calculated with regard to atmospheric drag or rolling resistance, not to mention variables in terrain that occur with actual road driving; all these factors will bring the number down to a more realistic level. Regardless, the working value is what we'll be concerned with for the remainder of testing; changes in it will be easily tracked and indicative of the impact our modifications have on overall consumption.
Now that our control and baseline numbers have been established, a more concise hypothesis can be made as to how we plan to drastically improve fuel economy. Considering our plans to reduce rotational and static mass, lean air/fuel ratios, and loosen restriction on airflow to and from combustion, we are shooting for a real-world 35mpg highway average. Adding forced induction at the final step, which will increase power production without significantly adding mass or increasing drag, we plan to elevate the real-world value to 45 mpg. We've set our sights high, but not unrealistically so. According to Edmunds.com, the Toyota Yaris is the most efficient gasoline-burning, conventional car (as opposed to a hybrid or an electric) on the road today. But even this vehicle achieves its estimated 40 mpg highway efficiency while still maintaining the traditional "margin of safety" around its tuning and build that we plan to eliminate in our project, not to mention it does so with natural aspiration-and we're betting properly tuned forced induction will help us even more. Can we accomplish our goal? If not, how close to it will we come? Can we surpass it, even? Most importantly, can our methods be adapted to your car, to increase its efficiency? More answers are coming in the next installment, when we dive into modification, to see just how far the OEM and aftermarket can help, along with the influence of a good tuner.