Changing Variables for Chemical mixing control

Hey Guys,
new member here. I am trying to figure out a solution for a chemical mixing system we have at work. Currently we are mixing a chemical and water, both of which are individually pumped into a mix tank through flow metered circuits. Currently the system works by controlling the valve open time on both circuits in order to provide a proportional mix as determined by the ratio determined in the HMI. The issue is that the value that is used to validate the chemical is within spec is the TDS. We have found that the TDS of the bulk chemical is fluctuating a significant amount from the manufacturer, causing a need to constantly change the proportions in the HMI when the system is supposed to be fully automated.

My theory is that instead of using proportional mixing, to instead go to a control based off the actual live value of the TDS with a EC probe in the mix tank. The water level will be controlled to a set point with an existing ultrasonic level sensor. Then chemical pump (flojet pneumatic diaphragm pump) will be controlled with a simple on/off control based on the TDS value being within tolerance or not similar to a PID controller. This way the system can compensate for the discrepancy in TDS from the manufacturer.

Does this make sense? Am I missing something here?
I am new to all this and would love to see other solutions.
Sounds like you want to make a control loop with PID control.
The total dissolved solids are being measured currently how? is this a fluctuation in the instrument or the actual product?
If it's the product ...then I would say that is the root cause is at the source. If that's not feasible to solve then yes a control loop will most likely sort this out. ....but maybe a proportional valve to control the flow....or an inverter to control some sort of pump would sit on your plant side of the control loop. Bang bang control sounds a bit ruff but I'm not sure what you are doing? this what you are proposing?? fig below
Probably a batch system will need to convert the PID to a timed pulse output if you are going down that route of bang-bang control on the PID side.
I would imagine you would do this mix on the fly and towards the end perform finer adjustment...but this should naturally happen if your PID is set up correctly...
Many ways to skin a cat....usually a process guy will have a deeper understanding and know the optimal way ( you would think )
but from a controls perspective, this is very doable and should not really present any major has been done many many times