Marek Lubieniecki's banner

Marek Lubieniecki

@m_lubieniecki • 2,146 subscribers

Simulation & optimisation engineer Robotics · Aerospace · Motorsport · AI

Shorts

2026 Monza won't be the flat-out Monza we know. Even in qualifying, with a full battery and only a 5 MJ/lap recharge limit, there won't be enough energy to sustain full power down the straights. Cars will deploy hard at the start of each straight, then slowly lose speed until the next braking zone. About 3 MJ of that 5 MJ can actually be recovered under braking - the rest has to come from on-throttle harvesting. Sim lap time: 1:23.53, +4.7s vs. 2025's pole. Exact number will depend on how much aero drag/downforce each team's setup carries. #F1Tech #ItalianGP

2026 Monza won't be the flat-out Monza we know. Even in qualifying, with a full battery and only a 5 MJ/lap recharge limit, there won't be enough energy to sustain full power down the straights. Cars will deploy hard at the start of each straight, then slowly lose speed until the next braking zone. About 3 MJ of that 5 MJ can actually be recovered under braking - the rest has to come from on-throttle harvesting. Sim lap time: 1:23.53, +4.7s vs. 2025's pole. Exact number will depend on how much aero drag/downforce each team's setup carries. #F1Tech #ItalianGP

420,923 views

After Monza, Madring is shaping up to be the exact opposite problem. Monza = legacy track, long straights, few corners, drag is king. Madring = brand new for 2026, barely any straights, 22 corners. I'm not worried about energy management here -with drivers constantly braking and on partial throttle through 22 corners a lap, the MGU-K stays fed naturally. The real challenge is the track model itself. It's not just tight and full of chicanes - it also has banking, and getting that right matters a lot for the optimal line. Preliminary qualifying sim: 1:27.26. I'll keep trying to improve it, but honestly hard to tell how close this is without FP1 data :) #F1 #F1Tech

212,114 views

Shanghai’s opening complex (T1–T4) is a tightening spiral where the car turns almost 270° before the exit. Balancing grip and power through this section is where drivers can make a real difference - and it’s also why it’s tricky to model. Here's how my model approaches it:

Shanghai’s opening complex (T1–T4) is a tightening spiral where the car turns almost 270° before the exit. Balancing grip and power through this section is where drivers can make a real difference - and it’s also why it’s tricky to model. Here's how my model approaches it:

1,027,888 views

How will 2026 cars handle Baku's 2.2 km straight? Better than you'd think. It's the only part of the lap that really demands electrical power. In qualifying the car arrives with a nearly full battery - 3.95 of 4 MJ in my sim - and dumps the lot before the line, crossing it at zero. The 90° low-speed corners aren't very downforce-dependent either, so the smaller 2026 cars aren't punished there the way they will be in fast corners. My lap: 1:42.53, about 1.4 s off Verstappen's 2025 pole. Other configs land +1.4 to +2.2 s - so, close. Two details to watch: the optimiser doesn't deploy into T15 at all, running the last 150 m to the braking zone on the engine alone to save the energy. And it takes a small lift at T18, 272 to 268 km/h through the kink. Whether real deployment maps behave that way, we'll see. I modelled the 8.5 MJ recharge limit; the lap didn't reach it, at 8.29 MJ. #F1 #F1Tech #AzerbaijanGP #MotorsportEngineering

How will 2026 cars handle Baku's 2.2 km straight? Better than you'd think. It's the only part of the lap that really demands electrical power. In qualifying the car arrives with a nearly full battery - 3.95 of 4 MJ in my sim - and dumps the lot before the line, crossing it at zero. The 90° low-speed corners aren't very downforce-dependent either, so the smaller 2026 cars aren't punished there the way they will be in fast corners. My lap: 1:42.53, about 1.4 s off Verstappen's 2025 pole. Other configs land +1.4 to +2.2 s - so, close. Two details to watch: the optimiser doesn't deploy into T15 at all, running the last 150 m to the braking zone on the engine alone to save the energy. And it takes a small lift at T18, 272 to 268 km/h through the kink. Whether real deployment maps behave that way, we'll see. I modelled the 8.5 MJ recharge limit; the lap didn't reach it, at 8.29 MJ. #F1 #F1Tech #AzerbaijanGP #MotorsportEngineering

72,800 views

What does Verstappen mean when he says Silverstone is now "constantly flat"? I ran the 2026 lap through my optimizer to find out - and it might back him up. From T5 to T15, you may not need to brake at all - if you time the MGU-K recovery to superclip and feed the battery. 🧵

What does Verstappen mean when he says Silverstone is now "constantly flat"? I ran the 2026 lap through my optimizer to find out - and it might back him up. From T5 to T15, you may not need to brake at all - if you time the MGU-K recovery to superclip and feed the battery. 🧵

347,511 views

To find the fastest lap around an F1 track, you wouldn't start with a 3D map of the circuit. Instead you reduce the whole thing to two numbers that change as you drive: how hard you're cornering, and how much room you have. 🧵

To find the fastest lap around an F1 track, you wouldn't start with a 3D map of the circuit. Instead you reduce the whole thing to two numbers that change as you drive: how hard you're cornering, and how much room you have. 🧵

229,897 views

Sepang is back on the F1 calendar after almost 10 years. The last pole lap there was Hamilton's 1:30.076 in 2017, when Mercedes dominated the second half of the season and Ferrari dropped the ball (it seems nothing has changed…) As far as I know, the qualifying harvest limit isn't confirmed yet, so I ran my 2026 sim from 5 to 9 MJ/lap: 5 MJ → 1:33.112 6 MJ → 1:32.618 8 MJ → 1:31.916 8.5 MJ → 1:31.807 9 MJ → 1:31.751 Going from 5 to 8 MJ is worth about 1.2 s. After 8 MJ the gains are marginal, just 0.165 s more up to 9 MJ. The extra energy is harvested harder into the braking zones and deployed for longer down the straights. (Forgot to run 7 MJ, sorry 😄) Treat this as a teaser: this model doesn't have all the improvements from Baku yet, and it only has the two straight-line-mode zones marked in green.

Sepang is back on the F1 calendar after almost 10 years. The last pole lap there was Hamilton's 1:30.076 in 2017, when Mercedes dominated the second half of the season and Ferrari dropped the ball (it seems nothing has changed…) As far as I know, the qualifying harvest limit isn't confirmed yet, so I ran my 2026 sim from 5 to 9 MJ/lap: 5 MJ → 1:33.112 6 MJ → 1:32.618 8 MJ → 1:31.916 8.5 MJ → 1:31.807 9 MJ → 1:31.751 Going from 5 to 8 MJ is worth about 1.2 s. After 8 MJ the gains are marginal, just 0.165 s more up to 9 MJ. The extra energy is harvested harder into the braking zones and deployed for longer down the straights. (Forgot to run 7 MJ, sorry 😄) Treat this as a teaser: this model doesn't have all the improvements from Baku yet, and it only has the two straight-line-mode zones marked in green.

20,937 views

Spa isn't a race track. It's a rollercoaster. 🎢 Every lap the car climbs to +54 m above the start line, then plunges ~100 m down to the lowest point of the circuit, 48 m below where it began. That elevation isn't scenery. It's shaping the lap, and in 2026 more than ever. 🧵

Spa isn't a race track. It's a rollercoaster. 🎢 Every lap the car climbs to +54 m above the start line, then plunges ~100 m down to the lowest point of the circuit, 48 m below where it began. That elevation isn't scenery. It's shaping the lap, and in 2026 more than ever. 🧵

106,969 views

Verstappen said Spa's sector 2 isn't fun in the 2026 cars - "F3 power" through there. So I simmed an LMP2 (~65 HP more than F1 fom ICE) against real F1 Q3 telemetry. At the end of the last straight, where the brakes go on, they arrive at nearly the same speed: 297 vs 292 km/h. #F1 #LMP2 #BelgianGP

Verstappen said Spa's sector 2 isn't fun in the 2026 cars - "F3 power" through there. So I simmed an LMP2 (~65 HP more than F1 fom ICE) against real F1 Q3 telemetry. At the end of the last straight, where the brakes go on, they arrive at nearly the same speed: 297 vs 292 km/h. #F1 #LMP2 #BelgianGP

95,805 views

We know the MGU-K harvests energy under braking. But how much can an F1 car actually recharge in one corner? Austria T3 is an interesting case: ~319 → 73 km/h, steeply uphill - and even here, the MGU-K recovers less than 30% of the car’s removed energy. 🧵

We know the MGU-K harvests energy under braking. But how much can an F1 car actually recharge in one corner? Austria T3 is an interesting case: ~319 → 73 km/h, steeply uphill - and even here, the MGU-K recovers less than 30% of the car’s removed energy. 🧵

79,315 views

The Shanghai circuit should be quite battery-friendly thanks to several heavy braking zones. But what about the long back straight? I ran my lap optimisation model on the Chinese GP layout to estimate the 2026 energy behaviour. Full results below. 👇 #F1 #ChineseGP

The Shanghai circuit should be quite battery-friendly thanks to several heavy braking zones. But what about the long back straight? I ran my lap optimisation model on the Chinese GP layout to estimate the 2026 energy behaviour. Full results below. 👇 #F1 #ChineseGP

140,997 views

You cannot simulate Zandvoort's Turn 13 correctly if you treat it as flat. Not "less accurate" - literally impossible. Real 2025 data: the car does 209 km/h through the apex, banked at 18°. Model the same corner flat and it only holds ~127 km/h - an 82 km/h gap from tilting the track alone. Zandvoort has two such corners, Turn 3 and Turn 13, both banked up to 18°. Two separate mechanisms create that gap - gravity pulling you toward the corner, and inertia getting redirected into the track by the tilt. 🧵

You cannot simulate Zandvoort's Turn 13 correctly if you treat it as flat. Not "less accurate" - literally impossible. Real 2025 data: the car does 209 km/h through the apex, banked at 18°. Model the same corner flat and it only holds ~127 km/h - an 82 km/h gap from tilting the track alone. Zandvoort has two such corners, Turn 3 and Turn 13, both banked up to 18°. Two separate mechanisms create that gap - gravity pulling you toward the corner, and inertia getting redirected into the track by the tilt. 🧵

28,834 views

The hardest lap I've thrown at my sim - my best shot at an energy-optimal 2026 Spa qualifying lap 👇 🔋 Full deploy up Kemmel - battery empty by Les Combes 🔌 Sector 2 mostly recharges, banking it all for one last dump down Blanchimont Not sure this one's spot-on though - there seem to be several roughly-equal strategies, so read it as an approximation, not the answer. 1:46.5, ~6s off the '25 pole. My drag's calibrated on Silverstone; teams trim wing + cooling for Spa's low-drag spec, which I don't model — so 4–5s still feels believable. #F1 #BelgianGP

The hardest lap I've thrown at my sim - my best shot at an energy-optimal 2026 Spa qualifying lap 👇 🔋 Full deploy up Kemmel - battery empty by Les Combes 🔌 Sector 2 mostly recharges, banking it all for one last dump down Blanchimont Not sure this one's spot-on though - there seem to be several roughly-equal strategies, so read it as an approximation, not the answer. 1:46.5, ~6s off the '25 pole. My drag's calibrated on Silverstone; teams trim wing + cooling for Spa's low-drag spec, which I don't model — so 4–5s still feels believable. #F1 #BelgianGP

34,004 views

Did you know we can see F1 engine RPM in onboard sound? A V6 fires 3x per crank rev, so 12,000 rpm becomes ~600 Hz in the audio. Track those bands and we can ask: is the ICE still loaded off-throttle in corners to charge the MGU-K? More details below 👇 Idea partly inspired by Heinz Harald Frentzen comment that the ICE now revs higher in corners. #f1

Did you know we can see F1 engine RPM in onboard sound? A V6 fires 3x per crank rev, so 12,000 rpm becomes ~600 Hz in the audio. Track those bands and we can ask: is the ICE still loaded off-throttle in corners to charge the MGU-K? More details below 👇 Idea partly inspired by Heinz Harald Frentzen comment that the ICE now revs higher in corners. #f1

26,009 views

During the Australian GP weekend we saw cars losing up to ~50 km/h at the end of the straights. What actually causes that? I ran a simplified 2026 hybrid model to explore the effect. Here are the results from the Australian GP lap. More details in the comments 👇#F1 #F1Tech

During the Australian GP weekend we saw cars losing up to ~50 km/h at the end of the straights. What actually causes that? I ran a simplified 2026 hybrid model to explore the effect. Here are the results from the Australian GP lap. More details in the comments 👇#F1 #F1Tech

44,621 views

Videos

Anya Rossi

sweetdream.ai

SweetDream.ai•Sponsored•Livecam

Watch Anya Live

Anya is streaming live right now! Join her private show and enjoy exclusive content.

Exclusive private shows

1.2k viewers online

Private Show

Join now for exclusive access

Free preview available • Premium content

Sepang 2026 qualifying prediction: 1:31.83 Setup: 7.5 MJ harvest limit, four Straight Mode zones (main straight, back straight, T3→T4, T8→T9). Where the 7.5 MJ comes from: 🔹 48% braking, spread evenly over T1, T4, T9, T14 and T15 (~0.6 MJ each) 🔹 52% on corner exits: the ICE runs hard, the tyres can't use it all yet, so the MGU-K takes the surplus 🔹 ~0% superclipping at the end of the straights Sepang doesn't need to superclip with 7.5 MJ and four zones. Instead the MGU-K switches off for the last 100–300 m of each straight: the car runs on the 400 kW ICE alone and slowly bleeds speed against drag. T5–T6: the model takes the esses flat at ~290 km/h, about 5.9 g. At Spa this year, cars took Raidillon and Blanchimont flat at ~300 km/h, so it's plausible, but it feels too fast. Caveats: ▪️ Track geometry is the TUM centreline; I checked the width against satellite imagery at T5–T6 and it's within ±1 m of the painted lines ▪️ The whole car stays inside the white lines, with no kerb use ▪️ In high-speed corners the model has been up to ~9% optimistic before (Pouhon at Spa) Let's see on Saturday.

Sepang 2026 qualifying prediction: 1:31.83 Setup: 7.5 MJ harvest limit, four Straight Mode zones (main straight, back straight, T3→T4, T8→T9). Where the 7.5 MJ comes from: 🔹 48% braking, spread evenly over T1, T4, T9, T14 and T15 (~0.6 MJ each) 🔹 52% on corner exits: the ICE runs hard, the tyres can't use it all yet, so the MGU-K takes the surplus 🔹 ~0% superclipping at the end of the straights Sepang doesn't need to superclip with 7.5 MJ and four zones. Instead the MGU-K switches off for the last 100–300 m of each straight: the car runs on the 400 kW ICE alone and slowly bleeds speed against drag. T5–T6: the model takes the esses flat at ~290 km/h, about 5.9 g. At Spa this year, cars took Raidillon and Blanchimont flat at ~300 km/h, so it's plausible, but it feels too fast. Caveats: ▪️ Track geometry is the TUM centreline; I checked the width against satellite imagery at T5–T6 and it's within ±1 m of the painted lines ▪️ The whole car stays inside the white lines, with no kerb use ▪️ In high-speed corners the model has been up to ~9% optimistic before (Pouhon at Spa) Let's see on Saturday.

Marek Lubieniecki

27,652 views • 12 days ago

No more content to load