Commuting - Determining 'true speed'

I've become a data junky with my commute.

I have a Garmin Etrex GPS on my handlebars.
I load all the tracks from my commutes in to Garmin Map Source software (to see few of all my routes), then use GPs Babel to convert it to GPX to load it in to GPS Action Replay (to run the 'dot race' of my past commutes, and see where slow and fast points are.)

The trip home from work almost always takes longer than the trip to work. Work is at a slight elevation decline from home (a couple hundred feet over 6 miles), the prevailing winds tend to blow to the office. The commute home is more lilely to be in the dark (thus, needing the hub-generator light on.) The traffic lights home cause more delays. (right turns at long lights there, with left turns returning.) And, I tend to be more tired in the evening.

The GPS can get elevation information and delays at lights. However, is there a good way to determine wind drag? Is there a wind gauge that could be strapped to the bike? (And then somehow calibrated with bike speed to get actual wind speed?) And then there is the matter of putting all of these things in to some sort of formula to determine exactly how hard I am pedalling. (maybe a heart rate monitor for the control?)

Any good ideas to satisfy the need to know just too much information about the daily commute?

I think you just a need a fistful of pitot tubes on your fork and an instrument kit from an F-18.

I think you are talking about a power meter. They tell you how many watts you are generating. They are also $$$$, but you don't seem to be too averse to that.

For an accurate speed, use a wheel magnet. A GPS's speed is based on the distance in latitude and longitude only. Anytime you go up or down a hill, the speed and distance are compromised.

For an accurate speed, use a wheel magnet. A GPS's speed is based on the distance in latitude and longitude only. Anytime you go up or down a hill, the speed and distance are compromised.

On anything less than a 20% grade, the difference is unlikely to be more than the accuracy of a computer. (+/- 2%)

On anything less than a 20% grade, the difference is unlikely to be more than the accuracy of a computer. (+/- 2%)

Maybe so. However, it seems to me that the OP is one who is looking for a measurement that is as accurate as possible. ;)

I think you are talking about a power meter. They tell you how many watts you are generating. They are also $$$$, but you don't seem to be too averse to that.
x2 ... who cares about distance ... it's all about the wattage per hour.

That is the truest indicator of effort.

The OP also said that he is using software to determine speed. The software has elevation data in it, and could potentially include that into the speed calculation. The gps data that he is keeping is probably points and time rather than points and speed. In either case, the computer could easily correct for elevation.

you could use a data logging wind speed gauge, ofcourse you also need to correct for variations in air density caused by fluctuations in barometric pressure, temperature, altitude, and humidity so you need to datalog those variables. As well as speed you also need relative wind direction because the drag factor varies with a crosswind, most of these variables can be datalogged with instrumentation used in yacht racing, any aerospace contractor would also be able to help you.

Ofcourse, carrying these instruments and their power source will add weight and aerodynamic drag so you'll have to develop a computer model that will correct for that before you can get your true performance. You'll need to establish some baseline performance measurements under controlled conditions to develop an accurate model, see if NASA can help, they have the wind tunnel and necessary instrumentation.

Let us know how you go.

The OP also said that he is using software to determine speed. The software has elevation data in it, and could potentially include that into the speed calculation. The gps data that he is keeping is probably points and time rather than points and speed. In either case, the computer could easily correct for elevation.

I think the key word in your statement is 'potentially'. Even if the computer's software does account for elevation, there are other issues that come into play, making a GPS a less than perfect speedometer.

Most modern GPS units do measure altitude, as long as there is reception from four or more satellites. However their software, (in the unit), does not take this altitude into account when calculating speed. This is why the Garmin unit that they recently released, specifically designed for use on a bicycle, uses a wheel magnet to calculate speed.

A hard wired speedometer that measures rolling distance will always be more accurate than a GPS. Civilian GPS units can only display your whereabouts within 3 meters. And to attain that much accuracy, the unit has to be WAAS-enabled and average a stationary reading for a minimum of 120 seconds. For maximum accuracy, using a GPS, the user would be stopping every 3 meters to let the GPS average its location. This doesn't even take into account reception problems that can and do arise from cloud cover and foliage.

Using a GPS to measure speed is fine for most recreational cyclists. But, as I mentioned above, it sounds like the OP wants something more accurate. :)

I agree that a good bike computer with logging would be better.
However, many people on the board cite a gps as being 10 ft or so inaccurate, and imply that means that they suck at calculating speed. That is very untrue. Successive measurements are not randomly off by 10 ft. If one measurement was off by 10ft to the north, then the next is likely to be the same. The speed is very stable.
For one reason or another, my gps is more accurate than my speedometer in my car. A bike computer should be better, but it does require some accurate measurements of tire diameter for calibration.
The real advantage of gps for the OP is that he wants the speed measurements to be stamped with a location, so that he can determine average speed as a function of position on his route to work. I'm not sure if this would be possible with a logging bike computer, but it would be more tedious at best and require writing a custom computer program. I think that a palm pilot (serial connection) would make a good logging platform for a wind meter, but I don't like this idea from the start.
Your commuting. Just try different routes, and keep track of commute times if you are trying to optimize that. After commuting for a while, most people don't seem preoccupied with shortening their commute, though. They just want it to be fun, interesting, and hassle free.

You could try mounting a small weather recorder like:

http://www.rei.com/online/store/ProductDisplay?storeId=8000&catalogId=40000008000&productId=47946279&parent_category_rn=4500570&vcat=REI_SEARCH

to your bike and download the "air speed" to your computer. You would need to merge it with your other data but it might do what you want.

you could use a data logging wind speed gauge, ofcourse you also need to correct for variations in air density caused by fluctuations in barometric pressure, temperature, altitude, and humidity so you need to datalog those variables. As well as speed you also need relative wind direction because the drag factor varies with a crosswind, most of these variables can be datalogged with instrumentation used in yacht racing, any aerospace contractor would also be able to help you.

Ofcourse, carrying these instruments and their power source will add weight and aerodynamic drag so you'll have to develop a computer model that will correct for that before you can get your true performance. You'll need to establish some baseline performance measurements under controlled conditions to develop an accurate model, see if NASA can help, they have the wind tunnel and necessary instrumentation.

Let us know how you go.

Cyclaholic is right. You also might want to put the measuring instruments somewhere in front of or above the bike (on a rod), since your body and bike will be pushing air and thus affecting the measurements. Getting useful information from the measurements will require some deal of (interesting?) computing, because of various nonlinearities in the effect of windvelocity, apparent windangle etc. on drag.

Regarding bikecomputer v.s. GPS units, I think that the major difference in accuracy of speed measurements are:
1) sampling rate. Bike computer samples once per rotation of wheel, but this is adjustable for GPS units (but I don't think that a normal GPS devices offer high enough sampling rate to outperfom the bike computer).
2) bike computer has to be calibrated to tiresize and pressure (no calibration needed for GPS) and quite many bike computers out there are not calibrated correctly (but those who wan't data with high accuracy probably do calibrate there bike computers carefully)

I agree that a good bike computer with logging would be better.
However, many people on the board cite a gps as being 10 ft or so inaccurate, and imply that means that they suck at calculating speed. That is very untrue.
...
I never said that a GPS receiver's speedometer sucks, I said that it is not the most accurate tool for measuring speed.



...
Successive measurements are not randomly off by 10 ft. If one measurement was off by 10ft to the north, then the next is likely to be the same. The speed is very stable.
...
Actually, the algorithms used to skew the data being received by a GPS unit generate a random number within the "accuracy window." As such, one measurement can be and often is thrown out in another direction from where the previous measurement was taken. (See Exersise #s 1 and 2)

In most cases, 10 ft here or there is not going to be noticeable. However, if you want the most accurate data, use something other than a GPS.

Exercise #1:
Display the "breadcrumb trail" as you walk or ride a winding path. When you get to the end of that path, turn around and retrace your steps. The return track will not follow the first. In fact, it won't stay consistent in its variations, (i.e.-it will crisscross the previous track, showing that the data has been randomized).

Exercise #2:
Ride or drive at a speed over 10 mph for a mile or so. As soon as you stop, set a waypoint for your current location. Without moving from where you recorded the waypoint, set it as your destination. Again, without moving, wait for 2 to 3 minutes and you'll find that your actual position is likely somewhere between 25 and 75 feet from where you set that point.

Don't get me wrong, I love my GPS and have it with me on most rides. I even display the unit's speedometer, sometimes. However, if I were recording the data for some purpose, other than my own enjoyment, I would want that data to be as accurate as possible. So I wouldn't rely on a GPS.