Collecting Power Data for IMLP
I am interested in collecting a pile of power data for IMLP. My interest is in tinkering with the data in search of the holy grail: correlation between power and bike split.
Yes, there are many complicating factors such as size, weight, aero-ness, gearing, weather, mechanical or bio breaks, etc. but I am still interested. What I have done so far is review all Race Reports I could find covering the past 3 years and collect any info that seemed potentially pertinent. This data is in a Google doc along with 2 quick charts to get a feel for the data. Unfortunalty the data are incomplete making it difficult to do much.
So what I am asking is for peeps with power data on IMLP to post some/all to the spreadsheet. Names are optional as is any data you either don't have or don't want to share. However, voids in the data will exclude that record from charts that utilize the missing data. Some data fields can be calculated from others (FTP = NP / IF).
Go here to add data: IMLP Power Data
Chart: NP-Time_IF
Based on NP and time in minutes. Colors indicate sex (M,F) and size is scaled by IF. Google excludes data points missing IF.
Chart: FTP-Time
Simple FTP vs time in minutes.
I would love to do more with weight, w/Kg, NP/AP etc but between limited data and the lack of functionality in Google Docs, I will need more data and do it offline in Excel.
Thanks!
Comments
Will you also include run split, and pre race FTP and VDOT in the data set? "There's no such thing as a good bike followed by a poor run"
I filled in my data, but feel free to toss it out if necessary, because my data is probably bad, but I figure thats a call for someone else to make
@ Coach P - how about some of your data? I only found 1 report on LP but you've done it like 8 times or something
@ Jeff - thanks
@ Al - That is very possible but my first focus is LP 'cause I am doing that this season. I wanted to keep the data to a single venue to have 1 less variable. I imagine power data from IMFL is very different than LP. I would be interested in collecting data, but it may benefit running this as a pilot project where I/we could determine minimum standards for data attributes.
As to the run, you are absolutely correct. An ideal data set would be free from any 'boogered' bike splits so all data would reflect well executed races. The thing is, getting around the course on a bike is really a physics experiment. Whether an athlete goes too hard on the bike and walks the 26.2, doesn't change the amount of power it took to cover 112 miles. Do you have any thoughts on how to use run data? filter on Good, Fair, Poor? some other more concrete metric?
@ Ryan - do you have a pwr file? Can we get some of the other metrics somehow?
@ Jim - me too on the w/Kg but I need more data to before we can really start to see things.
Meant to add what a great idea this is. Seems like this could be added to our TSS tables and really help out the newbies trying to figure out race day stuff.
I can already see a nice trend between w/kg & time - except for Jim D - hey Jim help an EN brother out - how the heck did you get such a great bike split
Here's a few additional random thoughts/opinions
I consider myself one of the FTP challenged ( < 3.0 w/kg) so i'm always looking for any advantage i can to make up for my limitations. I've ridden LP course 10x over last couple years and my take on the course is 1/3 uphill + 1/3 downhill + 1/3 flat. I ride a Cervelo P2, compact crank, 11/28, PT & wheel cover, Speedfill top & bottom. I consider my position pretty aero (although i don't know my drop) and I'm capable of riding aero the whole way. I've become a low VI jedi master thru alot of practice. I think I could ride 1.02/1.03 if i had 3 rings up front.
My bike split (7:14) included 12 min non riding - a few stops @ special needs, stretching, & a bike check
My thoughts on low VI @ LP since we all chase the elusive 1.0 in hopes of getting the best run.
For LP bike I used 2 data devices (910 & edge 800). One was set for counting zeros and the other not counting zeros. Data in your table is NOT counting. IMO hilly courses like LP with the 5 mi Keene desent (mostly coasting) and the big desent into Wilmington = a lot of zeros added and that plays havoc with your VI. My VI when counting zeros = 1.17......w/o zeros = 1.06
Since I'm FTP challenged I have a problem keeping my uphill watts in the correct range on several sections of the course (need more gears or higher FTP). On those sections I ride upright (non aero) knowing my upright FTP is several watts higher than my aero FTP (BTW the FTP in your table is my aero FTP). My feeling is I'm going slow enough uphill the advantage of aero savings are outweighed by minimizing my match burning using my higher upright FTP.
Following up on Al T thoughts - i think i rode LP about as good as i could given the fitness i own and the course requirements - leading me to believe i was pretty well setup for the run - on my run i adjusted my pace down for higher temps based on my HR - i effectively ran the whole way excepting about 1/2 mi - my run paces were surpisingly constant on both laps if you look @ the river rd section (flat) - again leading me to believe i had the best run i could given the conditions & my fitness level
Great project for the team - thx for your efforts
see u @ LP this summer
Jeff
I am wondering what power/weight ratio you need to go sub-6 hours Lake Placid. I don't see myself getting any better than 3.5 w/kg this year, and not sure if that's good enough. There is a hole in the data to know for sure - we need more data!
One use of something like this would be a way for someone with known FTP to get an estimate what his/her bike split will be, and therefore where to enter the TSS tables to get a target IF.
@ Jim - That was my ultimate goal: to find the power/weight ratio range at the 6 hour mark. No doubt the data will be noisy (aero, bio breaks, VI, weather, etc) but first look clearly shows a nice trend.
NEED MORE DATA!
If you have Race Rehearsal power details or actual Race data, I would love to get access to it. Either enter the data yourself at
IMLP Bike Splits
Or you could email me any powers file at swest262@gmail.com
Thanks to those who have already entered data!
Quick observation:
I went in to LP in 2012 with a slightly higher FTP (5 watts higher than 2010 - arguably a rounding error), but was about 30 minutest slower. It was hot and windy in 2012 (there were freakin' uphill winds!) so there are definitely other factors to consider.
All pre 2010 data is for a different course, the long out and back to Hazelton and where it was in the loop made for a very different ride.
The LP course offers a few places that present opportunities for different types of riders:
1) Obviously there are a lot of sustained climbs and how you approach them (we're assuming we all did them EN style) on the first loop will effect the second loop and the run.
2) The downhill to Keene can be approached cautiously or aggressively depending on the athlete's comfort level - several minutes can be gained or lost here.
3) The "flat section" from the bottom of Keen until the right hand turn uphill to Wilmington will reward folks that have their aero-fu dialed-in and punish the ones that don't.
Excellent idea. Thanks Cary.
Weather information (via WolframAlpha) for the appropriate year might provide insight as well:
Awesome! Thanks...
Tinkering with the data I was looking to normalize the data based on rider weight. Some data were missing weight so I guesstimated based on stature and sex. The result of this is below.
Based on the quick charts attached to the Google Doc, general correlations were evident as expected. My desire was to look at the data points relative to a trend line and try to find a way of describing the variations from the norm. First thoughts went to looking at FTP or W/Kg but then I realized it is average power that applies to getting weight X around the 112 miles, including the effects of gravity up and down hills in time Y. Stops for bio or mechanical were left in with the understanding that +/- time variability of several minutes were inevitable but routine.
Using average power and bike split times I charted them along with a linear trend line to visually see how the data looked. As expected several of the lightest riders were far from the trend line and had a significantly lower AP than others with similar bike splits. I assumed it was largely due to differences in weight. The problem was I had several data points without listed weights AND I had no clear way to use the weight in adjusting a user's power or time. How much different is one weight-time-AP from another? I decided to look at the average weight (AWt) of all points. That came in at 158-ish. I then assigned a +/- value for each weight I did have relative to the AWt. These +/- values were set as error bars on the Y scale (AP) in an arbitrary attempt to see how it worked. The results look convincing but I was still missing some weight data for a few. That is when I filled in those voids with calculations from FTP and W/Kg and estimates for a few.
So, thought?