I want to design an Airflow closed-loop control system. Now there is a question.

The following are devices in this system:
(1) Airflow analyzer with a velocity sensor: measure the sensor point airflow;
(2) MCU board: read the current airflow value from Airflow analyzer via RS232, and then perform a PID calculation, then control the PWM duty cycle of the FAN;
(3) FAN: has a programming PWM wire, PWM duty cycle change can reduce/increase FAN speed, if the FAN speed changes, the airflow(velocity) may change.

Some parameters of this system:
(1) FAN PWM frequency: 20kHz, duty cycle(50%~100%), the FAN cannot run if duty cycle<50%, it has a dead area.
(2) It takes about 10ms(Max=20ms) for MCU to read airflow data from Airflow analyzer through RS232 under Baud rate=19200b/s. The baudrate cannot be changed because Airflow analyzer's is fixed.
(3) MCU pid calculation will spend 10ms(some float operations).
(4) MCU has other housekeeping functions such like ADC button,LEDs display via SPI. These will spend several ms.

So the closed-loop sample&modulate time interval is about 40ms. The fan is 48Vdc/0.22A, I don't know this loop can acheive stable status. I mean if the PID's Kc,Ti,Td are selected properly, will the control system be stable?

Generally, if we do a Fan control,which is the suitable Sample Time interval?
Any guy has the experience?