Ironman Lanzarote Bike Disappointment - Post Race Analysis

[JoeX]

Hi all,

I’m very happy to have finished at all, but naturally I’m wondering what I could have done better after a disastrous bike split.

The Plan
I planned my effort on BestBikeSplit as below, an IF of 0.66 got me the optimum time for a reasonable cost in TSS.

6.5hrs looked a little optimistic to me.

The Race
The BestBikeSplit power file failed to give me target power after 8 minutes racing, so I was winging it based on my reading the crib sheet a few days before. I knew my rough targets for flats, in lines, climbs and descents.

The wind was much harder than predicted, I did a lot of climbing in aero position and generally felt good until I passed the 60km in 3 hours! This was 45mins late on my target. I stopped, thought there was a slight brake rub, corrected the brake and continued wondering what the hell had happened and when I would finish.

With 20km left to go of the 180, I caught a puncture and snapped the valve attaching the co2 canister. Realising I’d destroyed any chance of repair I sat on the roadside wondering how to get back to the hotel. Luckily, eventually, the course mechanics came and got me back on the road. I was over the moon but tried to keep my effort down so I didn’t ruin this second chance.

I felt good off the bike and started the run well. No aches, feeling well fed/hydrated.

Post Race
Even knowing my problems I was shocked to find that my bike split was 8h56. Even with moving time only, it was still 8h16.

I took the powerfile put it back into BBS for analysis, it reckoned I had only put in an IF of 0.62. So maybe I just didn’t push hard enough…except that I put this back into BBS and updated the weather data…it still thought I should get 6h55 for that effort.

So I’ve looked at in TR as below. I’m still not sure what I did wrong.

Split into roughly 3x 60km sections

Whole view

[TedP]

Best Bike Split obviously uses a ton of variables; if any of them are significantly off it can have a huge impact on the predicted time.

Couple of things come to mind:

1. Did you run your power file back through Aero Analyzer to get an updated CdA (sounds like you did when you said you “put it back into BBS for analysis”)?

2. Did you update your weight (with kit and all equipment), loaded bike weight (including nutrition and spares), component type, etc in the “My Bikes” tab? My understanding is that BBS uses the mechanical loss component from that page regardless of what you put as CdA.

3. Your CRR seems unrealistically low. Depending on tire and road conditions it could easily be double what you have input.

4. BBS produces an estimate based on an optimal pacing strategy. It seems like you unfortunately had to wing it based on the cue sheet so I imagine your pacing strategy was not exactly the same as what BBS suggested.

All that being said, it is still an estimate and modeling tool so I would still expect a difference.

[JoeX]

Thanks for your comments.

I ran the race file back through the race analyser, it estimated a World Class Time Triallist CdA so I’ve raised a support ticket.

I’ve not manually entered a cda or crr, they’re the standard calculated values off the values below. I’ve selected “Advanced Triathlete” because the FAQ says Advanced for “most triathletes with a bike fit”.

My pacing strategy but I can’t find a way to compare the plan to the race in that detail. Could it really account for 1h15?

[Michael_Tate]

@TedP

• Did you run your power file back through Aero Analyzer to get an updated CdA (sounds like you did when you said you “put it back into BBS for analysis”)?

• Your CRR seems unrealistically low. Depending on tire and road conditions it could easily be double what you have input.

These are the two that jump out at me. Especially with cdA on a windy course, since aerodynamic drag is proportional to the square of velocity, that begins to have a bigger actual impact on your predicted speed when you’re in nasty wind if your cdA is off. (BBS with 0 wind would be closer given error in your cdA than for an event with lots of wind). The Crr is also suspect.

Overall though it sounds like quite the nightmare

[JoeX]

The crr input was very similar to the values on bicyclerollingresistance but I’ve updated them to 0.00387 as per their review for gp4000s II at 100psi. Combined with Good (Smooth Asphalt) bbs calculates the value 0.00287. The roads were smooth for the most part, with a few very rough sections dotted about.

This is the Race Analyser output for what it’s worth, like I say I think there’s something wrong with the data as it estimates fantastical values from the file.

@Michael_Tate It was a tough ride but I felt better than on previous IM courses, to be honest. There were times I was looking at 160W and seemed to be barely moving

[baggiebird]

Couple of questions if I may hope you find them helpful.
How does your run compare with your normal runs. Quicker might mean you left some out there, slower would indicate a real tough day.
Your cadence is pretty low, is this normal or were you cooking your legs, I note the last third is down on Power.
I find BBS a real challenge, have you loaded sample rides in to get a good estimate, weighed your bike and yourself (fully tapered and ready to race), you will be a kg or so heavier, on a hilly windy course it dies have an impact.
What were your long rides at IF wise, how did this compare? An IF of 0.61 does feel a little low, it would demonstrate good power control uphill but maybe a little too much soft pedalling downhill.
Would be interested in your answers…

[TedP]

One thing I just thought of was the fact that you stopped and got off twice during the race. That will definitely skew the data (especially if your watch was still running) and may account for your bizarre climb CdA. Ideally you could use a file with no interruptions on a loop or out-back course to get an updated CdA value.

Bicycle Rolling Resistance uses a drum to measure rolling resistance (so basically ideal conditions). The tooltip from BBS indicates the CRR under “average road conditions.” I would probably suggest “poor” for road conditions based on that since you are using an already inflated CRR value.

I would take a look at the mechanical loss input at the bottom as well and set your component choice to “entry level.” Basically just set everything to the worst case/pessimistic scenario as it rarely works out as perfectly in the real world as it does in BBS.

[JoeX]

@baggiebird

Couple of questions if I may hope you find them helpful.

Thanks for chipping in

I simulate my race efforts over ~45km steeper hill rides at .63-.68 IF, this did feel easier but I put that down to the smoother gradients. Either way I’m okay that the power output was about ten watts off, and this accounts for ~30mins of the 1h45min time differential between the planned ~6:30 and moving time of ~8:15

My legs were absolutely fine, better than ever in an IM. This also squares with a 0.61 IF…but not the overall 313TSS. The TSS indicates I overbiked and I did struggle with dizziness after 1h35 on the run.

@TedP
Is it possible the IF is skewed due to the amount of time I was off the bike in the last 60km? Or does TR account for that.

It’ll take some time for me to find a relatively flat route I can use to generate a personal CdA so I’m happy to use a generic value that reflects my “okay, not stellar” position.

[JoeX]

[quote=“JoeX, post:5, topic:16502”].

[/quote]

Have I misunderstood the output of this?

Is the current value the value that was used until this analyser was run, and the estimated value is the “proposed” value? Or is it the other way around?

The crazy estimated values could explain the differential.

BBS have come back to me having analysed my race file but not answered the questions I posed. I was casting doubt on my ‘brilliant’ race CdA but I’m inferring from the below that the data backs it up and to only correct the climbing CdA.

“As for your climbing CdA- this is currently a known problem with our Yaw angle adjustment. Your CdA clusters look great as seen below, but obviously the average climbing CdA does not much the clusters in the graph:
​

​

As I mentioned we are aware of this and are looking into some improvements for it. For now, I recommend taking a rough estimate of the CdA clusters and using that as your estimated climbing CdA.”

[JoeX]

Okay I’ve been as pessimistic as possible, changing and bbs have emailed me CdA values of 0.2586 for race and 0.4569 for climbing, this changed the predicted finish time by a few minutes…

I updated the course with one from this year’s race, which changed a few minutes too…

I’m still at a predicted time of 7h20 when I got 8h16.

If I can’t get the BBS model to predict a similar outcome for a real event, then I can’t see how it has value for me. But I know other people get very close outcome/simulation so something else must be wrong.

How does BBS distinguish between pedal power meters and crank or wheel?

Is it possible there is something wrong with the actual bike?

It was in okay but not pristine condition after a year on the trainer. Degreased, wiped down and reIubed.

I noticed a bit of chain rub in the lowest gear, so I didn’t use that gear on race day.

I thought there might be brake rub in the first 40k but I got similar power distance over the first and second thirds of the course.

I can’t think of anything that might cost an hour.

[JulianM]

How does it compare to previous IM bike splits - in terms of time, average power, normalised power and the like? Take the BBS out of the picture for the moment - I do use it as well and it can be useful but I have to get creative with some of the settings to get it to match the times I ride for both triathlon and TT’s.

Without the BBS input what would have your expectations been given the profile and weather that is so significant in Lanza?

[JoeX]

@JulianM

Without the BBS input what would have your expectations been given the profile and weather that is so significant in Lanza?

It’s an interesting point Julian, I’m going to have to think that one through.

My last IM was 2017 and my first with a PM, but I suffered significant dehydration. I guess as my FTP was lower then the IF would be less significant than the NP…boy this is a lot more involved than it should be!

[JulianM]

The reason I ask is that Lanza is quite possibly the hardest IM bike course there is and it looks like conditions were a bit harder this year than the previous year. Looking at some of the splits from athletes who rode this year and last, even at the front of the AG fields they are often 10-15 minutes slower and the longer you are riding the more disproportionally you are affected. I try to keep an eye on my AG results and one guy who rode a 5hr bike split in Kona (in a normal weather year, not last year when times were stupidly quick all round!) rode just under 6hrs in Lanza. Obviously I’ve no idea whether he had other bike issues like you did but there are other riders whose times are significantly slower than I would expect.

A 5% difference in conditions for a 4hr rider can mean a 12m slower ride which translates as a 24m difference for a 8hr ride and that doesn’t take into account the fact that winds generally tend to increase as the day progresses which again will disproportionally affect those who are on the course for longer.

One other thing to keep in mind is that the PD curve and IF calculations are time based rather than distance based. If you look at the often quoted chart for expected TSS and IF for Ironman splits and translate that ‘good’ range for tops out at around 66/67% for a 6:30 ride. If you continue that charts progression to the 8 hour or so mark then and IF for a ride of that duration isn’t hugely unreasonable.

There may well be other factors at play as you have suggested but I’m trying to look at some of the other context around the day and what may or may not have contributed to a ‘disastrous’ ride! Power is great, as is other data and the predictions it can make but it’s you that needs to push the pedals around… sometimes life gives you lemons and it’s uphill and into the wind!

[Alex2204]

@JoeX I encounter the same disappointment with BBS:

I did a triathlon last year in the Pyrénées. For the Bike split, I used BBS to estimate my time. Previously, I tried to nail my Cda running their Race Analytic tools. Despite following the recommended target power, I ended with an extra 20’ minutes (4h vs 3h40).

After the race, I ran again the race Analytic tools to adjust my Cda, rerun the race plan with the exact NP of the race. It still gave me a 20’ minutes difference with my actual race time.

DId the same exercise this year and it got even worth: the time difference jump up to 30 minutes…

I asked BBS for help and the only answer was that I did not properly follow the power plan, which I can’t really .check - anyone has a tool to compare the power files ?.

One key element: this was in a very mountainous environment, meaning complex winds and snicky roads. So I can understand the difficulty to nail the data model… On other races, this estimates were closer to the actual time.

I don’t think it will help you but at least you know that you’re not alone…

[rjessop]

I used BBS for the first time at Holkham and it was eerily accurate overall but like Alex I don’t have a means of evaluating my adherence to the plan in detail. I came in slightly ahead of time, but I did overcook it a bit (slightly lower AP and higher NP, so higher Variability Index). Race Analytics is in alpha; maybe they’ve got a feature to compare performance to plan in the works already for the next phase?

Also in my case my estimated race CdA looks super low and my climb CdA looks super high (the average is close to the estimate from the bike page); which makes me think I’ve got my total weight (rider + bike) significantly out as I assume it’s looking at ascending and descending speed?

[TristanM]

Have you done any other races or test rides with BBS?
I use it for myself, and for my coached athletes, and whilst I find it is excellent if the inputs are correct - it is very much a case of ‘junk in, junk out’
I always do some validation rides after inputting the data, and have a screen with target power, actual power (3s ave), target speed and actual speed showing (as well as some other data).
What I find is that the pre-calculated cdA and cRR figures are often wrong initially and need to be tweaked. A solo ride in race kit, about 1hr can be useful to use to calculate, but you still may need to do some manual tweaks to the inputs.
The other thing to check on the pacing plan is the weather forecast. I always go in and update the race plan the day before the event with the latest weather advanced weather - from experience, I have seen that this has changed quite a bit in the days running up to events, so worth paying attention to.
Ultimately, you need to validate the pacing plan before using it is a race. I can rely on my BBS data to be within a comfortable margin of error (taking into account race dynamics and human infallibility). If you are holding the target power, but the target speed is off, then either your inputs are wrong for cdA or cRR, or the weather is different to the forecast in the race plan.
Finally - the old saying ‘no plan survives contact with the enemy’ is always true. Have a race plan, but make sure your RPE is calibrated to your race pace. That way if something fails or changes you have a manual backup.

[JoeX]

@TristanM

do some validation rides

This is definitely something I wanted to do, but looks like my next time on road is IM Switzerland in ten days. My plan is to aim a little higher but within normal parameters for an IM bike.

[JoeX]

I’ve come to suspect the tyres although they span freely when checking, I’m not 100% certain so I’ve got a bike mech looking at it today and so far he tends to agree. The clearance is very slim and some flexion in the frame or wheel under load could cause is his initial diagnosis.

There is signs of friction on the tyre which was new, and he’ll be checking the frame although I’ve raced this bike for years so may be hard to tell.

[TristanM]

You can do it with a training ride. Just input the route into BBS and enter your target TSS for the ride. It will then give you plan to follow.
I’ve done this very thing for my athletes with short 20-mile rides to get some data validation.
For ‘next level’ stuff, and if you ride within a group, create a duplicate bike and call this [xxx-group], then use the race analyser to calculate your cdA for a group environment. You can then even ride to BBS in a group fairly accurately - although this has a larger margin of error (due to the drafting effect). It’s a bit of fun.

[cartsman]

Yup, certainly looks like that tire has been rubbing and that would explain the performance. It’s a long old day when you’ve effectively got the brakes on the whole way! What width are they and what’s the frame?