1. Variable Frequency Drive Test Results
Anish Gaikwad Deepak Ramasubramanian
WECC LMTF Meeting Salt Lake City, UT 3 Oct 2017

2. VFD Testing - Aim and Motivation
At present, for dynamic simulation, power electronic load is represented as constant P and constant Q, with the value of P&Q decreased linearly to zero between two voltage levels.
It is possible that such a representation is unable to capture the true transient behavior of a variable frequency drive.
As a result, due to the increasing proliferation of these devices in the distribution system, it is important to understand the behavior of the drive and ensure improved model representation for stability studies.

3. VFD Testing
Lab tests are being conducted on drives from different manufacturers with voltage rating of 3 phase 480V and/or 240V (line-line) and power rating from 7.5 HP to 125 HP.
The load on the motor is presently variable torque (torque vs speed).

4. VFD Testing
The performance of the drive is measured for a 3ph voltage sag of various magnitude and duration such as:
Depth
Duration
0.95pu
120 cycles
0.9pu
0.85pu
0.8pu
0.75pu
0.7pu
8 cycles ⁞
0.5pu
120 cycles or until trip
0.2pu

5. Present Status
Preliminary tests have been conducted on three 30 HP drives from two manufacturers and one 60 HP drive.
Simultaneously, the results of the tests are being processed and compared to identify trends, patterns and distinguishing characteristics that will aid in model development.
The test data will be used to develop better characterization of VFDs for bulk power system stability studies.

6. Basic theory of operation of front end of the drive
Vac
Vdc
Input Current, Active Power
Output Current, Active Power
Vac
Vdc
Input Current, Active Power
Output Current, Active Power
Tick-tock
Input Current, Active Power
Output Current, Active Power
Auto restart enabled

7. Trip time for three drives tested
Different time characteristics for each drive
Low voltage sag doesn’t necessarily mean trip
Trip region

8. Transient response for rectangular voltage sag to 0.8pu
Constant power consumption after transient
Significant transient behavior on individual basis

9. Transient response for rectangular voltage sag to 0.7pu
Zero power consumption even if drive doesn’t trip
Voltage thresholds in present model set as 0.7pu and 0.4pu

10. Transient response for rectangular voltage sag to 0.6pu
Varied time thresholds across all drives tested
Voltage thresholds in present model set as 0.7pu and 0.4pu
Linear decrease in power of present model captures to some extent both power reduction due to trip and power reduction due to lower ac side voltage

11. Key Takeaways
In the immediate aftermath of the voltage sag, drive power consumption has transient response
In the existing model, this transient cannot be captured
Transient is more pronounced in real power
Drives operate typically close to unity power factor
Existing default trip settings for power electronics load are 0.7 pu (upper) and 0.4 pu (lower)
Initial tests indicate that drives may not trip even at 0.2 pu
The voltage level(s) at which the drive trips has an associated time component.
The existing model has a hard voltage dropout threshold
The out-of-box default settings and characteristics of drives of different makes, can be vastly different.
The test behavior is of an individual drive. Behavior of aggregate model could reduce the effect of the transients?

12. Is the present algebraic model sufficient?
Questions to ponder..
Is the present algebraic model sufficient?
Pro: Easy to parameterize
Con: Transients are not adequately captured
Is a piecewise algebraic model required?
Pro: Sufficiently easy to understand
Con: Could introduce numerical issues in simulation; could be tedious to implement in software
Is a one/two differential equation model better?
Pro: Easy to capture approximate transient behavior; easy to implement in software
Con: Could be tedious to construct the correct equation and parameterize it.
For drive connected loads of larger rating, is a detailed model required?