2006 Prius data analysis

[TheForce]

Since I have finally been able to get my can-view data captured and decoded I have decided to do some data analysis with my 2006 Prius. Unfortunately I don't know that much about excel or OpenOffice to create my own graphs yet so for now I'm leaving it up to YOU to sort through the data. I'm hoping someone will volunteer to make some graphs for me. I know Bob Wilson likes to make lots of graphs with lots of info in them. If your reading Bob feel free to do what ever you want with my data.


Here is my current fill up log in ODS and XLS format. ( Current as of 2009-08-08 )
http://jaygroh.dreamhost.com/priusch...a/PriusLog.ods
http://jaygroh.dreamhost.com/priusch...a/PriusLog.xls


Electric usage statistics

August 2008 to the end of the year.
http://jaygroh.dreamhost.com/priusch...a/wudn2008.ods
http://jaygroh.dreamhost.com/priusch...a/wudn2008.xls


Here are a few files from my trips to and from work. These are in CSV format. The naming of the files are in YearMonthDate-24hTimeStarted.csv. There are about 45 lines of data for every minute captured if you want to know how long the trip was.
http://jaygroh.dreamhost.com/priusch...ewdataCSV1.zip

Last few days of May 2008 in CSV format. Each file is a whole day of driving instead of a single trip. File name is YearMonthDay.csv
http://jaygroh.dreamhost.com/priusch...2008MayCSV.zip

June 2008 data
There are about 7 days of missing data because my screen broke and I could not record the data. This zip contains the RAW bin files. You will need to use my V4+ decoder software found here to convert the data to CSV.
http://jaygroh.dreamhost.com/priusch...a/2008June.zip

July 2008 data
July 16-17 are combined. July 31 I ran the tank down to %0 to get the E10 out of it.
http://jaygroh.dreamhost.com/priusch...ata/200807.zip

August 2008 data
Aug. 14th is where the Hymotion battery data starts.
http://jaygroh.dreamhost.com/priusch...ata/200808.zip

September 2008 data
I'm still on my first PHEV tank.
http://jaygroh.dreamhost.com/priusch...ata/200809.zip

October 2008 data
Firtst tank with PHEV ended on October 11th.
http://jaygroh.dreamhost.com/priusch...ata/200810.zip

December 2008 data
http://jaygroh.dreamhost.com/priusch...ata/200812.zip

January 2009 data
http://jaygroh.dreamhost.com/priusch...ata/200901.zip

February 2009 data
http://jaygroh.dreamhost.com/priusch...ata/200902.zip

March 2009 data
http://jaygroh.dreamhost.com/priusch...ata/200903.zip

April 2009 data
http://jaygroh.dreamhost.com/priusch...ata/200904.zip

May 2009 data
http://jaygroh.dreamhost.com/priusch...ata/200905.zip

June 2009 data
http://jaygroh.dreamhost.com/priusch...ata/200906.zip

July 2009 data
http://jaygroh.dreamhost.com/priusch...ata/200907.zip

August 2009 data Last month of logs.
http://jaygroh.dreamhost.com/priusch...ata/200908.zip

I will no longer be logging data unless I feel like logging certain situations. I will still be updating my fuel log when I fill the take up.

[Sheepdog]

I hope someone picks this up as I am interested in all thing Prius!

[TheForce]

Updated my fuel log and added can-view data for the last few days of May 2008.

[hobbit]

I was just doing a little noodling over some other data sets
last night, as an exercise in figuring out how to graph stuff
in excel -- normally I hate all things microsoft but that's what
the scantool runs under and it squirts out .csv files so I might
as well do the reduction right there. I made good progress,
although trying to determine any fine points between fuel burn
rate vs. output power is being sort of indeterminate. At least
I know how to apply simple formulas to new columns and control
what goes into graph axes now.
.
In your sets, what are the units for fuel flow? And where does
the "vvti" figure come from and what are its units?
.
_H*

[TheForce]

My first set of CSV files are in metric units. My last few days of May are in US units.

Fuel flow is from CAN message 0x38 in Liters/hr but converted to Gallons/hr. Not sure if this is the correct location for the data but I'm just going by what was originally written in Morris's decoder program and Vass's spreadsheet.

Same thing for VVTi. I got it from CAN message 0x526. The VVTi value I get is the same that is displayed in the can-view.

If you have a list of all the different data and its location in the CAN messages I would appreciate it if I could get a copy off of you. There are still some data points I would like to be able to graph out like IGN BTDC. I don't know if its included in the can-view dump because I don't have the location of the IGN information.

Also you don't have to use excel if you don't want to. I'm using OpenOffice Calc to do work with my CSV files. I've been MS office fee for many years now. Now I just have to work on getting MS OS free.

[JimboK]

Hobbit, Jay and I have been working together on this ... actually he more than I. The tentative formula he has for the mystery fuel flow units and/or location seems to be in the ballpark. This is a (MS Excel) chart from a segment of my commute from earlier in the week:



Using his gallons/hour calculation, I created an Excel formula to figure iMPG. The chart results are pretty consistent with my historical CAN-View observations of the RPM/MPH/iMPG relationship. I plan to cross-check the fuel flow formula tomorrow with CAN-View's displayed values.

Now, as he suggests, it would be peachy if you have a handle on where in the data stream he can find ignition timing. Now that I finally have my CV data capture working, I'd like to analyze, among other things, the various operating parameters that folks are using in SHM.

Jay, many thanks to you, BTW, for all your help on this!

[hobbit]

That's sort of what I'm after, too -- more or less steady-state
highway conditions > 41 MPH. Ignition timing can be queried,
whether or not you know where it is [or not] in the passively-
captured CAN chatter, it's just a bit slower data rate. I've
found that if I try to hold a given set of conditions for some
reasonable time [10 sec?] on the highway I can get reasonably
consistent captures that match each other. Often the data for
a given round of queries gets into weird edge conditions because
running parameters changed over the window when a scantool asked
for and received all the various parameters, but hopefully staring
at a scatter graph allows seeing the basic trends and visual
discarding of outliers.
.
To me, the important params are RPM [usually used as my X axis
in what I've been going over recently], injector time, measured
ICE torque, and then maybe veh. speed and ignition timing. The
first three allow calculation of power and fuel-consumption
RATE, i.e. integrating injector times over RPM should give
a figure that can be matched against (torque*RPM) to yield power
and let time cancel out. But I'm not sure I'm calculating things
quite right. I should probably convert everything to SI and
take a wild-ass guess as to "energy flow rate" based on roughly
10 KWh or so of actual car-pushing energy available per gallon
of fuel...
.
There was also of course some recent stuff on the electrical
side exploring boost voltage, but that's a totally different
intent and approach.
.
_H*

[bwilson4web]

Sorry I haven't been able to help but let me offer a couple of excel tips:

1. split "Y" axis scales - often we'll have more than one data values that when mapped on one scale make the other data look flat or throw it off the graph. To solve this problem, 'click' to select a graph line and then use the pulldown "Format" for "Selected Data Series" and choose "Axis" and select "Secondary axis." This will plot that data series on a new axis scale on the right. You can adjust it to have both sets of data displayed. I often use a special format like "lines and point" for data on one side and just "lines" on the other.
2. insert gaps in graph - many times it is useful to have a 'gap' in a line to make it clear something changed. Just insert a blank "row" and a gap will appear in the line.
3. careful about sampling errors - often some data points, say MG1 torque, can change very quickly and even polling ICE rpm adjacent to the MG1 torque can lead to 'impossible' values. For very dynamic data, consider 3-point averaging to reduce the effect. It can not be eliminated, just made "less bad."

Bob Wilson

[TheForce]

I added my data for June 2008 to the first post. There are about 7 days of data missing this month because of a screen failure. These files are in RAW format and need converted with my V4+ decoder software. Links are in my first post. All following data will be posted in RAW bin format so others can convert the data as they like.

[TheForce]

I added my data for July. July 16-17 are combined in to one file. July 31 I ran the tank to %0 to try to get as much E10 out of it as possible.

I also updated my spread sheets for new tank data.