Summary: Exploiting acoustic streaming effects for microfluidic devices has been proven to be important for cell, microparticle and fluid manipulation in many fields such as, biomedical engineering, medical diagnostic devices, cell studies and chemistry. Acoustic streaming is used in acoustofluidic systems for directing and sorting microparticles as well as mixing and pumping fluids. To understand the underlying physics of such acoustofluidic systems and thus use them more efficiently in practical setups, computational modelling is critically needed. Although some work has been done to numerically model acoustofluidic systems, there are few studies to evaluate the capability and accuracy of different numerical schemes for analysing this complex multi-physics problem and to be directly validated by experiments. This paper aims to investigate the acoustic streaming effects caused by surface acoustic waves in a microchannel flow by using two different computational approaches to model the acoustic effects in three dimensions. In the first approach, we model the whole acoustic field caused by the oscillating lower wall. Here, the acoustic streaming effects were directly calculated from the density and velocity fields caused by the acoustic field. In the second approach, a low fidelity model is employed to capture the effects of acoustic streaming without modelling the acoustic field itself. In this approach, we substituted the velocity of a one-dimensional attenuating wave in the acoustic streaming force formula, and calculated the acoustic streaming force without using the density and velocity caused by the acoustic field. Both the computational methods are then validated by the results obtained from microflow experiments. The results from the second approach are in reasonable agreement with experiments while being more efficient in terms of computational cost. On the contrary, the first approach, while being computationally more expensive, allows to estimate the pressure field resulting from acoustic waves and thus predicts the dynamic behaviour of microparticles more accurately. Results suggest that the first approach is best to use for analysing the mechanism of microparticle and fluid manipulation in microfluidic devices.