Physics.Simulate accurate with high force

When I use physics.simulate to calculate end position of my gameobject it shows correct position only when enough force is given. If I push my object with middle or strong force end position showed by physics. simulate is very accurate, but when I push my gameobject only a little physics.simulate shows position too far, like I would give higher force.
RollerMove Pushes gameobject, checkposition checks end position, simulatephysics invokes.

 private void simulatephysics()

    {      

        touchhandler.listofcoll.Clear();

        touchhandler.hitramp = false;

        touchhandler.force = (float)Math.Pow(distanceToRoller, 3);

        touchhandler.RollerMove();

        endpos = touchhandler.CheckPosition(rollerobj.GetComponent<Rigidbody>());

      

            destination.transform.position = endpos;

            markeractivated = true;

      

    }

  public void RollerMove()

    {

        if (force > 100f)

        {

            force = 100f;

        }

        transform.GetComponent<Rigidbody>().AddForce(raydir *force* 15f , ForceMode.Force);

    }

    public float timeCheck = 8f;

    public Vector3 CheckPosition(Rigidbody defaultRb)

    {

        Physics.autoSimulation = false;

        defaultRb = GetComponent<Rigidbody>();

        Vector3 defaultPos = defaultRb.position;

        Quaternion defaultRot = defaultRb.rotation;

        float timeInSec = timeCheck;

        while (timeInSec >= 0)

        {

           timeInSec -= 0.05f;

           Physics.Simulate(Time.fixedDeltaTime);

        }//end while

        Vector3 futurePos = defaultRb.position;

        defaultRb.velocity = Vector3.zero;

        defaultRb.angularVelocity = Vector3.zero;

        defaultRb.transform.position = defaultPos;

        defaultRb.transform.rotation = defaultRot;

        Physics.autoSimulation = true;

        return futurePos;

    }

Note that this call isn’t some special case, it’s what is called normally during fixed-update if you’re not manually simulating so the results you get will be identical to that.

The only odd thing above is that you’re counting how many “Time.fixedDeltaTime” simulation steps to use with a down-counter constant of 0.05. That’ll obviously only work if the two are the same. I can only presume this is an isolated scene with its local-physics?