Back-to-Back (Dyno) Testbench and Bidirectional supply

[Blackgar]

I want to build a back-to-back testbench where 2 PMSM 5kw will be used. The second one will be used as generator to simulate a variable load for the first motor .

Both motors will share a common DC bus and a bidirectional supply will be used. I already own a 5kW bidirectional supply, the IT6015C-80-450 from ITECH.

There are cases where energy will be transferred back to the supply and my main concern is if this energy is handled internally (lost on the internal resistive load of IT6015C) or going back to the grid?

In case that energy going back to the grid, it will be quite complex cause I will have to speak with the electric power company.

[NiHaoMike]

In case that energy going back to the grid, it will be quite complex cause I will have to speak with the electric power company.

If there's no substantial energy source independent from the grid such as a battery, there will be no sustained export condition.

[f4eru]

it will be quite complex cause I will have to speak with the electric power company.

Probably not.
Not sure about your country, but most often a small export is tolerated, and the power company even does not notice it.
Also, as you put a common DC bus, a brake chopper is enough to dissipate the bit of braking energy when slowing down. Often inverters integrate a brake chopper already.

[max_torque]

With a "Back to back" setup, assuming identical EUT and "Dyno" eMachines, then the power supply is only making up the losses in most cases (see later!)  Typically you could consider a design point at around 25 to 30% of the power of one eMachine. ie you could test a 100kW eMachine with a 30kW supply.

The load case where the DC link supply must absorb power rather than generate power is when the rotating inertia of the rig is transfered back to electrical energy as the rotational speed falls, and here the power is of course proportional to the rate of change of rotational speed.

Three options are possible:

1) limit negative speed transients to those of a magnitude where the peak power created (KE to EE) is below the overal system loss at that speed, ie you never see a net output power from the rig, any KE turned into EE gets burned in the normal losses. This can work, but leaves no capability for either an emergency stop situation where you probably want to stop the rig from rotating ASAP, or for doing any full dynamic simulation/ transient work therefore:

2) use an additional mechanical friction brake system to Estop the rig, which can be triggered on a DC link over voltage. I've used both mountain bike and motorcycle brake components to do this, depending on the size of the eMachines being tested. Normally i have a spring ON, air pressure OFF type arrangement for fail safe (make sure a lack of air pressure to the cylinder holding the  brake off also triggers STO!)

3) Use a simple resistive load dump on the DC link, again, can be triggered by a DC link over voltage. This can be pretty simply built with a contactor or large Thyristor / IGBT (depending on how you want it to work). I've used laser cut stainless steel sheets to make a poor-mans high pulse power resistor before too, used to crow bar a DC link to close to zero volts to prevent more expensive electronics from being damaged by transient DC link over votlage pulses. Can be somewhat sacrifical if you only expect it to work under fault or emergency conditions etc