The sim can produce some pretty big thermals. Here you can see the air being pulled down a shaded hill to an area being heated by the sun, which causes the air to suddenly rise:
8 Likes
Let’s say “for now”.
Asobo giveth and Asobo taketh away. 
4 Likes
Amazing stuff!
I have toggle CFD using Shift + \ by default however I can only get it to appear by changing the menu option instead…any suggestions. (DX12 GTX 1080Ti)
Based on the menu description for the toggle key, it only works when using a helicopter.
Unreal! This is what we’ve needed! That looks amazing…I hope it wasn’t a fluke and flying through storms/hurricanes/etc is like this now. Can’t wait to try it out!
3 Likes
Was able to do more testing to confirm the checkbox’s for enable/reinjecting weather CFD actually are ignored, the FPS issue when first testing was also likely due to live weather server issues as I could not repeat that on further tests either.
I hope that’s correct, because soaring with only updrafts and no downdrafts, while a sailplane pilot’s dream, would be terribly unrealistic.
I’m a sailplane pilot, and I can tell you that Asobo hasn’t even seen a picky user until we start scrutinizing SU 11. Sailplane pilots understand and exploit every subtle nuance of atmospheric movement, from convection, to orthographic (ridge) lift, to wave. It’s going to be a little challenging regardless because there’s no way to replicate the seat of the pants sensations that are such a part of soaring, but we’ll definitely know if the simulated atmosphere behaves correctly or not.
Cross country flights will depend upon convective lift and sink both being modeled, because varying airspeed to spend less time in sink and more time in lift is key to a successful flight.
3 Likes
Like I said they definitely exist. I got pulled into a nasty one between two strong rising thermals and barely made it.
I’d really recommend you get into it with one of the freeware gliders (I use the Discus 2c), the more experienced glider pilots trying it the better to help them tune and improve it. It’s a very powerful system judging from what you can see when you turn on the devmode visualization.
3 Likes
Yup, there is definitely falling air. There’s screenshots in this thread showing the air rising and fading under various scenarios:
1 Like
It says that but it is incorrect. Try the skyhawk g1000 for example.
Agree, downdrafts exists 
I thought it was missing because those were not visualised but that thing is to help find updrafts i bet
1 Like
So playing around with the visualizations some more, it looks like the weather CFD is running in 2 parts, there’s a 20km cube that seems to calculate air flow over landmass (hills, valleys, mountains,etc) and thermals, then a smaller roughly 1km radius that takes into account smaller objects like buildings, then it blends these together.
The end result is amazing, flown a few airports I frequented back in the day where creeks/ravines/buildings on short final required you to be prepared to make quick power adjustments or come in a little higher to offset the inevitable drop once you came out of the updraft and the sim mimics those perfectly now.
7 Likes
Does the CFD show any effect at the cloud boundary and/or within clouds?
Our friend lknfly1 shared a quite harrowing, but quite epic, account of turbulence while in the clouds. Was this separate cloud turbulence or thermals feeding into where clouds also happen to be at that moment?
And feel free to throw up some YouTube videos so that those of us not in the Beta can see what fun we’re missing! 
1 Like
I’ll play around later to see if the local flow visualization shows anything at the boundaries. Larger cumulus clouds will definitely show significant updrafts below them though.
1 Like
I’ve tested some today and could really notice more updrafts and downdrafts in dense clouds.
The thing that is missing in my opinion in those visuals is that it should show different colours depending on the strength/velocity of those shown thermals. maybe it’s made like that to not be too easy to find good thermals. Will make it a bit more unpredictable even that visualization is active.
5 Likes
I agree. It’d be a great educational tool to see where they’re at their strongest and weakest.
1 Like
CFD from the environment perspective appears to be very good. However, I still struggle to understand the seeming ‘need’ of the end user for CFD to be implimented in aircraft flight models. Probably an unpopular opinion, but there seems to be a belief that the latest thing is necessarily the best and I do not think that it is currently anywhere near good enough. From a developer’s perspective, I am frustrated by the fact that three of the four major structures of an aircraft cannot be defined accurately:
- Fuselage - we are able to define the diameter, the length, the centre point and the ‘softness’. So in geometrical terms, we have a cylinder. How many aircraft fuselages are actually perfectly cylindrical when you move away from airliners?
- Wing - No definition of the airfoil section. Sweep appears to still be causing problems despite being highlighted very early on. There is absolutely no ability to define anything other than a basic wing - tailless deltas do not work, there is no stability (worked in FSX, MSFS without a tailplane gives you continuous backflips) Biplanes (or even triplanes) cannot be defined as there can only be one wing.
- Horizontal stabiliser - always perfectly level, no anhedral (e.g. BAe Hawk) or dihedral (e.g. Vickers Viscount) can be defined.
The aim of CFD is admirable, but we do not currently have the ability to define an aircraft anywhere near accurately enough, given Asobo’s own statement in the SDK that " The new flight model for Microsoft Flight Simulator relies on the shape of the aircraft to predict its aerodynamic behavior and we have almost entirely dropped the use of tables of data. Because of this, the correct definition of the aircraft’s dimension data is of particularly importance." That ‘shape’ which is of great importance cannot be defined accurately enough, and when the behaviours do not give the required behaviour we have no fallback because the ‘tables of data’ have been done away with.
I have been working recently on the PA38, notorious for it’s spin behaviour. With correct geometry and a flight model built using the manufacturer’s manuals, the older MSFS flight model could be made to spin reasonably accurately. Put CFD in the mix, with (as stated) an accurately defiined shape and it struggles to stall, let alone spin. Sure, I could adjust the geometry and probably artificially build in the required behaviours by using incorrect figures, but that will impact elsewhere on the flight model and lead to incorrect behaviours in other flight regimes. The whole thing soon becomes a guesswork mess, rather than a clearly defined and understood definition.
It’s not a problem if people want a ‘game’, but when you are aiming for accuracy across the entire flight envelope (with ‘simulator’ as your primary stimulus) it is a big issue. As for help from Asobo through the official channels, I asked back in July what we should do if the accurate geometry does not give the required / documented behaviours. The continuing silence makes me think ‘smoke and mirrors’. . .
18 Likes
I think your technical appraisal is very helpful and hopefully is being or will be considered.
However this emotional dismissal (‘It’s only a game until everything is modeled to my satisfaction, at which point I will graciously bestow upon it the title of ‘simulator’’), is really not very meaningful. Certain aspects of the clouds need work. Certain aspects of the terrain need work. Certain aspects of the the FM mechanisms need work. Certain aspects of this new CFD implementation need work.
And all the time, it is still simulating the total experience of flight, no matter how completely it models your own particular priority.
9 Likes