I know I’ve been a bit quiet this month and haven’t posted many new articles.  But I think this one will make up for that.  I’ve been pretty busy writing this tool in a form that I could share on here with everyone.

This is my expedition fuel planning tool.

Background

Planning for fuel usage can be kind of tricky.  Different conditions, the choice of fuel type, the type of pot used, how windy it is each day, altitude, etc all have an effect on the fuel consumption of your stove.

I’ve been playing around with my set up, doing some experiments.  I then used the data from those experiments to create a model of fuel usage that I could use to predict how much fuel I would probably need on a trip of a given duration.

This model is what is known as a Monte Carlo simulation. It runs many many scenarios given a set of data, and then outputs the average result of those scenarios.

How to use the tool

To use the tool, you will need to do two sets of experiments.  You’ll notice I say “repeat several times” for each experiment.  How many is “several”?  Well, the more the better.  At the most, 30 iterations.  Empirically, 30 iterations will give us a normally distributed sample, no matter what the underlying distribution looks like.  In reality, I’d say 5 or 6 runs should be plenty.

Experiment 1

Boil 1 liter of water using your stove, pot, and fuel in the way you intend to use it on the trip.  Time how long it takes to reach a boil.  Repeat several times.

Calculate the mean and standard deviation (MS Excel will do this for you quite easily).

Experiment 2

Measure a specific amount of fuel into your fuel bottle, and boil 1 liter of water.  Time how long it takes for that water to boil.  Measure the fuel remaining.  Repeat several times.

Calculate the mean and standard deviation of how much fuel your stove uses per minute.  Fluid ounces burned / minutes to boil is the formula you’ll use.

Data

Those four numbers you calculated (mean boil time, SD of boil time, mean fuel usage, and SD of fuel usage) will be entered under “experimental data”.

Then, you select the type of fuel used (for now, I have only entered white gas and kerosene. . . if you really want to use this for unleaded auto gas or jet fuel, I can add that easily enough).

Finally, enter the volume of water you plan to boil each day, and the length of your trip in days.

Output

For now, I output the mean, standard deviation, and mean plus two SDs in fluid ounces of fuel.  I also calculate the mass of the fuel for the mean plus two SDs.

Two SDs results in a pretty conservative number in terms of probable fuel requirements.  Lots of industries like to use six SDs, and you can easily calculate that yourself, or, if enough people ask, I can add that calculation to the tool.  You’ll have to decide your risk tolerance for yourself.

Precision and Accuracy

This tool is written in PHP, and the results are similar to the results I get when using MS Excel for the calculations.  I used the mt_rand function in PHP, as the documentation claims that it is a better random number generator than the rand function.  I didn’t actually plot the results from mt_rand to see the distribution.  I wrote all the statistical procedures myself (even for the mean and standard deviation calculations), so they depend on good inputs (e.g. real random numbers and solid experimental data).

Conclusion

I’d really like to hear feedback on this tool.  With some more users playing with it, perhaps it can be made even better.

Tags: , , , , , , ,

Leave a Reply

You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>