# Physics Engine Behaviour bug **URL:** \ **Category:** Bugs\ **Created:** [April 22, 2026, 11:04pm UTC](https://forum.babylonjs.com/t/physics-engine-behaviour-bug/63250 "2026-04-22T23:04:01Z")\ **Posts on this page:** 9\ **Page:** 1
**Author:** ![Leonardo\_Villalobos](https://sea2.discourse-cdn.com/flex024/user_avatar/forum.babylonjs.com/leonardo_villalobos/32/9628_2.png) [@Leonardo\_Villalobos](https://forum.babylonjs.com/u/Leonardo_Villalobos)\ **Post date:** [April 22, 2026, 11:04pm UTC](https://forum.babylonjs.com/t/physics-engine-behaviour-bug/63250/1 "2026-04-22T23:04:01Z")
The Babylon official Playground [Physics V2 Helpers | Babylon.js Playground](https://playground.babylonjs.com/#E5URLZ#20) shows the bug I was asking to check in question formulated somedays ago. The mentioned playground, using Edge for windows, when loaded from start or reloaded (not the scene restarted), act like the cubes are all moved suddenly by a force or an impule, but if you reset the scene you will see that is not the case, then why is it? In my case, the temporary work around was to set the bodies type of motion to STATIC and later change the to DYNAMIC. Can someone may explain what to do to avoid the situation?, without the workaround? ---
**Author:** ![Cedric](https://sea2.discourse-cdn.com/flex024/user_avatar/forum.babylonjs.com/cedric/32/2742_2.png) [@Cedric](https://forum.babylonjs.com/u/Cedric)\ **Post date:** [April 23, 2026, 7:29am UTC](https://forum.babylonjs.com/t/physics-engine-behaviour-bug/63250/2 "2026-04-23T07:29:38Z")
> However, how come that after restarting the scene, the effect is avoided? Persistent datas that do not need to be initialized again. ---
**Author:** ![Leonardo\_Villalobos](https://sea2.discourse-cdn.com/flex024/user_avatar/forum.babylonjs.com/leonardo_villalobos/32/9628_2.png) [@Leonardo\_Villalobos](https://forum.babylonjs.com/u/Leonardo_Villalobos)\ **Post date:** [April 23, 2026, 12:50pm UTC](https://forum.babylonjs.com/t/physics-engine-behaviour-bug/63250/3 "2026-04-23T12:50:32Z")
I understand what you suggest. However, how come that after restarting the scene, the effect is avoided? Isn’t it a bug? Are you planning to check? EDIT: Not even a comment about it? ---
**Author:** ![Leonardo\_Villalobos](https://sea2.discourse-cdn.com/flex024/user_avatar/forum.babylonjs.com/leonardo_villalobos/32/9628_2.png) [@Leonardo\_Villalobos](https://forum.babylonjs.com/u/Leonardo_Villalobos)\ **Post date:** [April 27, 2026, 12:49pm UTC](https://forum.babylonjs.com/t/physics-engine-behaviour-bug/63250/4 "2026-04-27T12:49:38Z")
So, no solution? ---
**Author:** ![Cedric](https://sea2.discourse-cdn.com/flex024/user_avatar/forum.babylonjs.com/cedric/32/2742_2.png) [@Cedric](https://forum.babylonjs.com/u/Cedric)\ **Post date:** [April 27, 2026, 1:02pm UTC](https://forum.babylonjs.com/t/physics-engine-behaviour-bug/63250/5 "2026-04-27T13:02:28Z")
This is not a bug. This is an effect of cache that can be mitigated with solutions provided earlier. If this solution doesn’t fix your problem, provide a PR. I’ll be glad to review it. EDIT: I don’t see my post that I posted a few days ago. Anyway, It listed a few options like this 1: > [@Havok - wake up from sleep mode](https://forum.babylonjs.com/t/havok-wake-up-from-sleep-mode/40852/6): > > Oh sorry! I totally missed this last week; was pulled into some projects which took a bunch of my time. The upcoming version of the WASM will have additional functions for sleep control, but looks like you’ve found a bug; changing some of the constraint params don’t send an activation request to the attached bodies. I’ll be sure to get this fixed up too, but Raggar’s workaround should hold you over 'til then. Thanks Raggar! Or use a fixed time step. Ot the solution to swith to dynamic bodies when scene is ready. Basically, anything to not have the duration time spike at the frame update time. ---
**Author:** ![Leonardo\_Villalobos](https://sea2.discourse-cdn.com/flex024/user_avatar/forum.babylonjs.com/leonardo_villalobos/32/9628_2.png) [@Leonardo\_Villalobos](https://forum.babylonjs.com/u/Leonardo_Villalobos)\ **Post date:** [April 27, 2026, 1:07pm UTC](https://forum.babylonjs.com/t/physics-engine-behaviour-bug/63250/6 "2026-04-27T13:07:17Z")
The PR with the bug explained is in your PG [Physics V2 Helpers | Babylon.js Playground](https://playground.babylonjs.com/#E5URLZ#20). It’s a bug beacuse it repeat itself when loaded or reloaded (F5), but the behaviour does not happen if you reset the scene. ---
**Author:** ![Cedric](https://sea2.discourse-cdn.com/flex024/user_avatar/forum.babylonjs.com/cedric/32/2742_2.png) [@Cedric](https://forum.babylonjs.com/u/Cedric)\ **Post date:** [April 27, 2026, 2:03pm UTC](https://forum.babylonjs.com/t/physics-engine-behaviour-bug/63250/7 "2026-04-27T14:03:31Z")
I did the smallest and easiest proposed fix in this PG : [https://playground.babylonjs.com/#E5URLZ#48](https://playground.babylonjs.com/#E5URLZ#48) ---
**Author:** ![Leonardo\_Villalobos](https://sea2.discourse-cdn.com/flex024/user_avatar/forum.babylonjs.com/leonardo_villalobos/32/9628_2.png) [@Leonardo\_Villalobos](https://forum.babylonjs.com/u/Leonardo_Villalobos)\ **Post date:** [April 27, 2026, 2:47pm UTC](https://forum.babylonjs.com/t/physics-engine-behaviour-bug/63250/8 "2026-04-27T14:47:27Z")
The only difference between the 2 PR is the HAVOC parameter set to false (constructor BABYLON.HavokPlugin(\_useDeltaForWorldStep?: boolean, hpInjection?: any, parameters?: BABYLON.HavokPluginParameters): BABYLON.HavokPlugin). However I’m using the UMD version of the plugin, and my problem persist. Is there any difference in behaviour between the UMD version and the regular version of the Havok plugin? ---
**Author:** ![Cedric](https://sea2.discourse-cdn.com/flex024/user_avatar/forum.babylonjs.com/cedric/32/2742_2.png) [@Cedric](https://forum.babylonjs.com/u/Cedric)\ **Post date:** [April 28, 2026, 7:38am UTC](https://forum.babylonjs.com/t/physics-engine-behaviour-bug/63250/9 "2026-04-28T07:38:08Z")
No difference. It should behave exactly the same.