Thorpe Park: General Discussion

[flyingguitar]

The ride now has a reputation for poor reliability.
Mainly because it isn't actually reliable...flawed by design.

I have been talking about the issue with weather, people have been talking about how bad it is when it really isn't that much of a problem, I don't see it down due to weather too often

however I would agree that the ride is too short and probably could have done with a block brake and double the length, and it has been problematic with its lift hill issues.

 

[Smiley]

you really haven't ever had to run a simulation.

I was running simulation and software validation on Sparc 10s as far back as 1994. Also an recognised expert @ ETSI for telecoms simulation, testing and validation. Tested/validated the most important software/hardware modules on satellites, that AFAIK are still working today, 20 years later in zero gravity travelling a bit faster than Hyperia! Not CFD, but not trivial. Perhaps that's why telecoms systems have MTBF measured in decades. An i9 with a GPU for 10 mins isn't exactly what I was thinking of - maybe 100 or 1000 running for weeks/months Like I said, you just rent your required compute from AWS or similar.

 

[Smiley]

An extremely good question!!

It’s clear Thorpe have upgraded their WiFi recently - there are hundreds of big white access points everywhere you look.

Not really sure why you can't do SCADA over wi-fi, 5G or microwave if you don't want to lay fibre.

 

[DistortAMG]

I was going to say, an i9 and an RTX 3080, isnt exactly cutting edge or efficient for running large and detailed simulations. Not with clould and AWS being so prominent. As someone said earlier. Renting cloud compute is ridiculously cheap.

Technology that companies such as Mack, would have access too, and would use for projects such as Hyperia.

@flyingguitar your arguments with simulation have just proven even more that there are design flaws with Hyperia, as companies do have access to huge amounts of clould compute for extremely precise sinulations.

 

[Dave]

So finally got around to riding Hyperia for the first time yesterday. Some thoughts:

No two ways about it, it’s a good coaster, the airtime and hang time mix is really good. The climb to the top of the first drop is suspenseful, and the twisty first drop is good which was surprising as I normally think twisty first drops are a bit naff. It’s also good in that once it’s warmed up even the cheap seats (middle rows) are fun to ride (was lucky to ride up front, the back and the middle over 4 goes).

But and it’s a big but, it’s half a roller coaster. When you consider Stealth and Oblivion which are short coasters they have a “feature” (launch/ drop) that is the signature delicacy that partly justifies the shortness (I do actually think both could have been better for being longer but that’s another story), but Hyperia doesn’t have one big tasty morsel to justify the ridiculously short layout.

Is it the best coaster in the UK? Maybe. The competition isn’t huge and the other competitor (Nemesis) has a very different dynamic. Is it better than Voltron as many people said? Hell No! Voltron poops all over Hyperia.

Headline conclusion, good but just as it was reaching greatness it stopped.

As an aside if anyone can explain why the wings need a shower during the dispatch sequence in the station let me know.

 

[flyingguitar]

I was running simulation and software validation on Sparc 10s as far back as 1994. Also an recognised expert @ ETSI for telecoms simulation, testing and validation. Tested/validated the most important software/hardware modules on satellites, that AFAIK are still working today, 20 years later in zero gravity travelling a bit faster than Hyperia! Not CFD, but not trivial. Perhaps that's why telecoms systems have MTBF measured in decades. An i9 with a GPU for 10 mins isn't exactly what I was thinking of - maybe 100 or 1000 running for weeks/months Like I said, you just rent your required compute from AWS or similar.

simulating software is quite easy I am not sure on what you have simulated so won't really comment further but considering how you have been going on as if simulations would have caught this, I am not sure they are that similar to the more engineering based simulations, when you are taught about simulations you are told that these are estimations based on inputs.

I was going to say, an i9 and an RTX 3080, isnt exactly cutting edge or efficient for running large and detailed simulations. Not with clould and AWS being so prominent. As someone said earlier. Renting cloud compute is ridiculously cheap.

Technology that companies such as Mack, would have access too, and would use for projects such as Hyperia.

@flyingguitar your arguments with simulation have just proven even more that there are design flaws with Hyperia, as companies do have access to huge amounts of clould compute for extremely precise sinulations.

no it isn't cutting edge, but I was trying to say about how costly and expensive it can be, that was an EXTREMLY simplified model, you are also missing my original point of the fact that there are assumptions to be made. also you can say about how 1 simulation won't cost much (even though it probably would cost quite a bit, there are cloud computing simulation software and it costs about £3-6 to get back one very basic simulation (an FEA)) but when designing parts, this simulation may have to be run hundreds of times, you can easily see 3,000,000 elements+ running with small time steps to calculate certain parts, which over low time steps can easily make extreme compute requirements

mack rides could do real time simulations with stuff like no limits 2, but that is quite simplified, the lifts are assumed to be at a constant speed, the brakes don't take account of brake pressure, application time etc, the drag/friction is just a constant multiplier each one of these assumptions are assumptions and just that, they can calculate the estimated deceleration, and calculating the drag but you would then be assuming it will be constant throughout the ride, it also doesn't take into account the flexing of the chassis adding energy losses, the wheel material deforming, etc

those additional parts could be simulated / calculated for each time step, but it can start dramatically increasing the simulation time, some parts probably won't need to be simulated (the brake simulations as a part of movement for example) but where you draw the line isn't easy, do you simulate fluids for each step (very computationally heavy!!) probably not, you would probably simulate it separately and find drag coefficients and do a very quick calculation to determine its effects, do you simulate the chassis flex effect on energy loss, probably not as it may require complex simulations, adding significant time and cost. it is a balancing act

you also have to realise it is impossible to simulate anything without assumptions (in the engineering world), CFD you have turbulence models to estimate how turbulence may be as it is essentially random, FEA you are often assuming the material is perfect with no flaws, etc. often to make simulations match they have to be calibrated (depends on the type of simulation, and how it is done), there are other assumptions you may not think of, we may be assuming the wheels are perfectly round, assuming the components are perfectly made, etc

essentially simulations are as good as the information you tell it, sometimes there are unknowns that are impossible to know and how they will affect the ride (normally they will add a compensating factor, but that is just a factor and may be too low or too high, or they may be perfect for normal senarios but not as good for odd scenarios, etc)