Useful simulations

[Z]

Back to USEFUL SIMULATIONS.
=========================

Claude,

I agree completely with your opening post. (Yes, really, and the world still turns! )

And also with Edward's, and susequent, posts.

But what do the students say? Where are your views on "simulations'? Specifically, how often can the students say they have made really USEFUL DECISIONS based on simulations?

I would like to hear examples of such, because I see precious few.
~o0o~

To get the ball rolling, consider the big-picture "point-simulators" that are very useful in setting the overall direction of a team. This direction includes deciding how much of the team's scarce resources, such as man-hours, money, +++, to spend on the different events (eg. gaining static-points vs dynamic), and what overall concept of car to build to maximise total points (eg. cost-points vs speed-points), and so on.

This type of simulation was used by Geoff and his team at RMIT over ten years ago, and was very successful. They were top-of-the-world for a few years around 2006+. Geoff then spelled out this approach in his "Reasoning..." thread, which most older-heads here recommend as compulsory reading for all FS-ers. Geoff even pointed out how easy it was to write his first simple point-simulator. It took ONE WEEKEND!

Then in 2012 Claude's company released a somewhat more sophisticted simulator (IIRC, written by Pete Ringwood?). This was aimed largely at the FS/FSAE world, and was/is GIVEN AWAY FOR FREE! Students need only google "OptimumLap" and find free access to track maps, typical FS-car parameters such as masses, hps, CdAs, ClAs, +++ (well, a lot of that stuff was there last I looked).

Sure, both the above are "only" point-mass simulators, but they are certainly enough to get a team started on the big-picture issues.

So what have teams done in the five years since OL was made FREELY available to all FS-ers on the planet?

Well..., following Julian's suggestion to "call them out personally", I present here three case studies taken from the very small FSAE-Oz world.
~o0o~

1. RMIT(C) - As noted above, around 2006+ RMIT adopted the "simple, lightweight, single-cylinder-450cc, 10" wheels, and no-aero" car-concept. They were the first team worldwide to achieve high success with this concept. This concept direction, of course, was encouraged by their own in-house, very simple, one-weekend-to-write, points-simulator.

Some years later (possibly after OptimumLap was freely available to all teams?), the later generation of RMIT students decided to CHANGE DIRECTION on overall concept. The single-cylinder engine was tossed and replaced with a larger capacity twin. And a TURBO was fitted. And a shed-load of carbonfibre was also added, with the car somehow managing to get heavier each year. The car was clearly evolving in the direction of "Mini-F1" (though not very mini on mass!).

However, despite the awesomeness of its new powerplant, no significant aero was added. Note that even a very simple simulator suggests that the main justification of mega-horsepower is to offset the extra aero-drag from a mega-downforce package.

Anyway, the end result of above changes is that RMIT cars have struggled to turn a wheel at most competitions in recent years.

So, can anyone at RMIT please explain how they arrived at their current car-concept, and what, if any, "simulator results" they have to justify this change?

Or, putting it another way, was this change driven purely by your testicles?
~o0o~

2. UTAS - In 2014 the Tasmanian team returned after a long absence from competition. Their 2014 car was of ~2006 RMIT style, being a quite simple single-cylinder car, albeit, and understandably, of relatively "inexperienced" execution (ie. many small deficiencies, and it weighed ~220 kg). Nevertheless, despite barely completing Skidpad and Accel, and not even starting AutoX, they finished just below mid-field at the 2014-Oz-comp (14/24 and 428 points).

If, for the next year, they just aimed to score points in all events, and perhaps tidied-up the car's design and fixed some of its defects (perhaps easily cutting 50 kg), then they would have comfortably made top-five in 2015.

But what did they do? They COMPLETELY CHANGED DIRECTION!

Their 2015 car was a ~250 kg heavyweight, with "awesome" four-cylinder powerplant, and all-options-ticked-including-the-kitchen-sink. Err, except for aero-package, which was planned but not delivered in time. And by the end of the 2015-comp they had gone backwards, with an overall score of 415 points.

In 2016 they piled even more junk on the car (yes, even a pneumatic-paddle-shifter for the gearbox!), and with a massive expenditure of time and money they managed to move a little ahead on points. But 2016 was the year where almost all the usual Top-Teams, even Monash, managed to shoot themselves in the foot.

By my reckoning (ie. my simulations), the very expensive 2016-UTAS car is barely faster dynamically than a tidied up version of their 2014 car (Edit: which had ~25 hp!). A better "engineered" version of the 2014 car (ie. better designed, developed, tested...) would be much faster. And a lot cheaper (and easier) to build, so better Cost score. And more Fuel Efficient.

So, can anyone at UTAS justify the above change in direction on the basis of "rational engineering decision making". Perhaps, something based on NUMBERS? Or is this yet another example of young men's (boys?) testicles taking control?
~o0o~

3. U of Melbourne - Very briefly, they moved from a four-cylinder spaceframe car in 2015, and SECOND PLACE OUTRIGHT, to a carbon-tub (new), and 500+cc-single (new), WITH A TURBO (new!), for 2016. And, no surprise, a glitch visited the engine/turbo and they scored ZERO DYNAMIC POINTS.

So, perhaps some justifiable changes on the overall concept front. But can anyone at UoM explain why their points-simulator insisted that their new concept, which entailed many big and potentially risky changes, MUST have a turbo?

How is it that your Sim-Guru missed these few lines of code in the points-sim?

IF (turbo-craps-itself) THEN
......LOSE ~500 to 625 points
...ELSE % assuming turbo ok.
......GAIN trifle points % from awesomeness of turbo-power.
......AND LOSE smidgeon points % from higher Cost and Fuel-use.
ENDIF

Based on the above addition to the Sim (perhaps done with pencil and paper), then, in the event of engine problems arising just before comp, the obvious action would be to implement THE WELL PREPARED BACKUP PLAN of "toss the turbo" (ie. turboless exhaust pipes built, non-turbo fuel map ready...).

I recall Geoff discussing their attempts to fit a turbo to their 2006+ car. Something along the lines of,
"...we found the optimum length of the turbo manifolds to be ... with the car at Silverstone ... and the turbo in the rubbish-bin back in Melbourne..."!

Z

[tromoly]

I recall Geoff discussing their attempts to fit a turbo to their 2006+ car. Something along the lines of,
"...we found the optimum length of the turbo manifolds to be ... with the car at Silverstone ... and the turbo in the rubbish-bin back in Melbourne..."!

I remember that quote, it came from this (http://www.fsae.com/forums/showthrea...ll=1#post25822) post:

Basically, we found that the critical dimension was the distance between the turbo body and the exhaust port - the further the better. For example, the car performed at its best when we were competing in Bruntingthorpe in the UK, while the turbo was located in a bin in Melbourne, Australia.

[Charles Kaneb]

Claude,

I think I made myself hard to understand there. I said that it would be a lot easier to make shims that gave a good, known adjustment than to make those that gave a bad, random adjustment!

My old teams didn't just calculate (and later measure) the effect of each shim, we chose sheetmetal thicknesses to make it easier to remember that "three shims on the LCAs give one degree negative camber".

[Z]

Tromoly,

Thanks for that, I couldn't find it from my quick search. Geoff said it much better!
~o0o~

Here are Geoff's thoughts on useful Simulations from his "Reasoning..." thread.

http://www.fsae.com/forums/showthrea...l=1#post117994

Some quotes from that post:

...there are two distinct categories of simulation :
1. Simulations that you learn about at uni
2. Simulations that are useful
... a useful litmus test for the value of a simulation is to show it to your university lecturer. The more excited he gets about it, the less likely it will be of any practical use to anyone.

- During the conceptual design phase, shoot anyone who uses the word “optimization”. During the detail design phase, do the same.

...for teams embarking on a new design project...
- If anyone mentions the words “transient”, “absolute” or “non-linear”, quietly walk them outside and show them the guy who mentioned the word “optimization”.

The secret to winning:
Convince your own team to simplify.
Convince your opponents they have over-simplified.

Z

[DougMilliken]

This is sounding a lot like the well known quote from George E. P. Box, "All models are wrong but some are useful".

Longer versions on the same topic can be found here, https://en.wikipedia.org/wiki/All_models_are_wrong It traces the history back to 1947 to polymath John von Neumann, "truth … is much too complicated to allow anything but approximations."

An interesting restatement by Box, "Remember that all models are wrong; the practical question is how wrong do they have to be to not be useful."

In RCVD and also in Learn & Compete, we introduce the Ladder of Abstraction, a mental construct to keep track of your current location between hard reality (a real car) and pure abstraction (perhaps a simple point-mass model of a car). It's near the beginning of RCVD Chapter 5, and Chapter 9 in L&C.

[Z]

Further to above quotes by Geoff, and specifically about "simulation for optimisation".

I recall a team-member from one of the teams I mentioned in my long post above, who a few years ago was crowing about how he had used the sophisticated FEA software his Uni provided, and that his lecturer was no doubt very excited about (!), to shave NNN grams from their previous year's machined aluminium rear-bulkhead. This was supposedly very impressive because it reduced the mass of said bulkhead by MM%. Yes, double figures of percent reduction!

And when finished, the whole car weighed ... ~250 kg!

BAD, BAD, BAD design.

Z

[Claude Rouelle]

Somewhat related, if not to simulations, at least to record you will need to create and carefully keep when you test not only to get fruitful tests but also to validate your simulations.

Also, sometimes there are so many inputs that you can't really qualify and quantify precisely. That makes your simulation neither useful nor relevant. In fact sometimes the best simulation you can have is the analysis of existing prerecorded data. That is why it is important to keep the most accurate and detailed test log book.

This is extracted (with some additions) from a post I made on March 14th this year

******

Test, Test, Test

(....)

When I ask students how much test they performed before the competition 80 % of the time I have very vague answer such as “x days” and when I ask them how many hours, how many laps, how many km they really spent on the test track running, I don't get an answer.

Even worse most of them do not have any written record of what did happen during these tests; some don’t even log the number laps ran.

Here is a bit of advice for testing planning and report As design judge, I would like to see test report that includes at the minimum
- Date
- Weather condition and noticeable change during the day. Atmospheric pressure, humidity, wind speed and direction.... A 300 $ weather station could be useful
- Track and air temperature before and after each run.
- Starting setup. You cannot be too detailed.
- What is the test plan: what are we doing today? You rarely won’t achieve all the test goals but you cannot start a test day without a plan. And that is the problem; I know too many FS teams that go to the test track with the goal of “running the car” with no other details…
- Cold and hot tire pressure, cold and hot tire temperature (these measurements should be systematic every time the car leaves and come back to the pits)
- Brake temperature
- Brake bias position: if the driver forgot to tell the team and this data is logged by a position sensor, you could make stupid mistakes
- same for ARB if they are adjustable by the driver from the cockpit
- Time at which the car left and came back to the pits
- Number of laps per run
- Each lap time
- Driver subjective feedback
- Setup change
- Careful note the exact amount of fuel added.
- Tire wear (you need to learn how to clean the tires from the marble and picked up rubber before you measure the remaining compound thickness)
- After the test an engineering report that combines driver subjective data and car objective logged data analysis, the setup evolution
- A Set Down. back on the setup at at the end of the day to measure what the setup exactly is
- Conclusions: what went well / what did not go well / why / how can we improve / next action plan. This part of the report should include the car failure analysis and the way the team work together

Claude

[Claude Rouelle]

The kind of info you should have on your setup sheet

- Date, location
- Student who is responsible (response-able) for the setup and who will sign it
- Description on how the setup and set down are performed
with or without driver (if so which driver because different driver will most probably mean different corner weight)
with or without fuel
which tire, which pressure (see more below)
which damper pressure (that could change your ride heights)
etc…
- Ride heights at each corner (describe clearly in a document for the team where and how the ride heights are measured)
- Number of shims corresponding to the targeted setup camber
- Number of shims (or pushrod/pullrod length) corresponding to the targeted setup ride heights
- Number of shims (or toe link length) corresponding to the targeted setup toe
- Corner weight
- Amount of fuel
- Front and rear caster, camber, toe
- Bump steer. A setup sheet MUST include bump steer numbers
- Spring and ARB stiffness (you need to measure those at the workshop do not trust what is written on the spring box or calculated from your FEA)
- ARB setting (if adjustable)
- ARB droop link length (if you are not careful you could introduce serious asymmetry in the setup, to be measured during the assembly phase and during the setup
- Spring preload
- Eye to eye damper fully extended length (double check those when you assemble your car)
- In fact, ideally, before the test (set up) and after the test (set down) every suspension linkage rod end center to rod end center length (you need to create a special tool to do that): toe link, pushrod, wishbone, ARB drooplink…
- Damper internal (type of piston, bump and rebound shims, etc..)
- Damper pressure
- Damper low and high speed, bump and rebound setting (“clicks”) if they are adjustable - I hope they are
- Wing angle, flap position (if adjustable - I hope they are) gurney flap type and size
- Specific suspension pick up points (in case they are adjustable - I hope they are)
- Pressure at which the tires were inflated to make the setup (usually the targeted hot pressure) –
- Tire set number used for the setup (unless you use dummy wheels for the setup); the tires you use for the setup is not necessarily the tire you use to test the car
- Tire number. You will use more than one set of tires so you better have keep a record of the tire set number
- Number of laps or Km already ran by these tires
- Rims part number and type (be careful that all rims have the same offset)
- Pedal box, steering wheel and seat position (in case those are adjustable)
- Type of brake discs and brake pads. Number of laps or Km already ran by those. Part number
- Tire cold and targeted hot pressure
- Differential setup (ramps angle, number of useful plates, preload (and make sure you know the difference between cold and hot preload)
- Master cylinder diameters
- Brake bias position
- Any useful engine and transmission spec (specific ECU setting for example)
- Radiator and brake duct opening or closing (if it is cold, people sometimes use tape to reduce the brake duct or water cooler radiator inlet surface)
- Also make sure you zero your suspension linear potentiometers, pushrod strain gauge
- Seat belt length per driver (so you do not need to adjust those at the track; you do that ahead of time at the workshop
- Seat, headrest used (in case they are different from one drive to another)

I hope this could help some of you

This is the kind of info we share step by step with more depth and with examples, spreadsheets, pictures, videos, stories of do and don't, stories of failures and success in the OptimumG Data Driven seminar