There is a very old paper( S.D.T. Robertson and K.M. Hebbar, 'A digital model for 3-phase induction machines' , IEEE Trans, on Power App. and Systems, Vol. PAS 88, 1969, p 1624, 29.) which might be useful . It is based on harmonic analysis of the rotor and stator fields .So 1.Obtain the individual fields over theta 2. Find Fourier series decomposition of both fields 3.Torque is given by  interaction of harmonics  Hope it is useful.

Cheers

There are three types of torque ripple:

1) Cogging, this a torque depending on position, without current.

2) Position dependent for a given current (should be vector in sine emf, constant emf in BLDC).

3) Ripple torque introduced by current ripple.

1) The cogging is often the most disturbing one. It can be analysed by considering magnets as current sources, Those current sources produce a torque ripple with the slot gaps or with teeth saturation. It can be analysed considering the the torque of a current source and gap combination (simple finite elements). To reduce the torque ripple all edges of magnets should have a different position, equally spread, compared to the gaps. Practical motors do get torque ripple by the the excentricity of the rotor. Motors without slots do not have cogging torques.

2) Often sine wave emf is used in combination with BLDC, it results in 15% p-p torque ripple as the emf is not constant. In sine wave control, harmonics generate torque ripples. The classical methods do predict rather well this emf.

3) The PWM is high frequent and produces rather acustic noise. Depending on the type of current control in BLDC the sum of currents gets dips during commutation between phases. Those dips generate also a torque ripple. .