Radar level would work to measure volume but bubbles and surface turbulence may make the level measurement noisy. A Coriolis meter in a recirculation line could measure density.  A pH electrode may not last long and may respond more to water concentration. I would use a high temperature glass electrode and a replaceable junction or solid reference electrode since these can better withstand chemical attack. There is the possibility that the use of a conductivity, oxidation reduction potential, and/or a pH measurement could provide an inferential measurement of concentration. I suggest you contact Jim Gray, manager of applications at Rosemount Analytical Irvine, Calif., to check application suitability (Editor's note: Jim Gray offers his insight below). Make sure all of the wetted materials of construction for the meter and the electrodes including o-rings and gaskets meet your worst case conditions considering the corrosive nature of the fluids in your process.
For more on radar level measurement, see the Control Global February 2012 Control Talk Column "Radar Love" and the Chemical Processing July 2011 article "Make the Most Out of Radar"

Expert Greg McMillan asked Ram G. Ramchandran of Systems Research Intl. of Destrehan, La., to provide additional inputs:

Here is Ram's answer: Anhydrous ammonia is corrosive in pipelines and some amount of water is injected with ammonia liquid in pipelines making it conductive with about 1,000 ppm [0.10%] or below water content. Using sonic probes/meters for level measurement is common. A sonic level transmitter should have dual probes with compatible metal and sheath and isolators. Another option is a capacitance probes with COMAD [RF] technology with all polytetrafluoroethylene insulated probes. I do not know his flange arrangement in the reactor. Extended diaphragm d/p can be tried with compatible metal facing on the flange face. A radar gauge will not lend itself well to this service. Generally tantalum sheath is used for thermowells to prevent corrosion. Some vendors offer a special instrument designed for this type of reactor.

A turbine meter is common for measuring flow. For custody transfer, nowadays, Coriolis meters are used in conjunction with inline provers. Operating temperature is -20°F, most of the wetted parts are 316 stainless steel. NO copper parts or wiring/housing is admissible in the operating area with the presence of ammonia gas.

If the converter uses electrical heaters to achieve exothermic reaction to sustain conversion- after the startup [catalyst reduction], usually a multi stream chromatograph is used to measure all gases, including inert gases like argon, at different points. These reactors operate at higher pressures and temperatures than converters common in the US. They are less efficient comparatively but less maintenance/failure-prone with longer catalyst life.

No pH is measured in converter area. Generally solid-state probes are used for pH. Besides CO, CO2, methane, O2 are measured using infrared analyzers. Conductivity is commonly used to maintain water quality in all exchangers. Continuous emission monitors [CEMs] are common on stacks for reporting NOx, CO, CO2 and O2.

Jim Gray, applications manager at Rosemount Analytical, offered this insight:

If you look at the conductivities of these solutions, they appear to be relatively straightforward concentration measurements:
21% ammonia has a conductivity of about 12,000 uS/cm and the curve of conductivity versus concentration always increases over this range.
21% ammonium nitrate likewise always increases and has a conductivity of 250,000 uS/cm at 21%.
The conductivity of phosphoric acid does increase and come to a maximum of 280,000 uS/cm at 50%, and then drops to 160,000 uS/cm at 70%.  Since there is a hump in the conductivity versus concentration curve, the conductivity at 70% would be the same as it is at 25%, which could cause some problems if they were relying on conductivity alone, depending upon what the process is doing.  Phosphoric acid solutions have a pretty healthy increase in specific gravity, increasing up to 1.25 at 40%, which is where my data runs out.
So, I think that if they are only interested in concentration measurements, a toroidal conductivity measurements with PEEK conductivity sensor, and maybe a density measurement for phosphoric acid would work.
I took a look online to see if I could get an idea of what the process might entail and it’s likely that they are making diammonium phosphate by reacting phosphoric acid with ammonia.  Since this is an acid neutralization reaction, pH comes to mind first.  But depending on the concentration of phosphoric acid they are starting with, there could be problems with the pH sensor.  If there will be exposure to concentrated ammonia solutions, there could also be problems because ammonia has a stronger affinity for silver ion than chloride does, which could cause problem with the reference.
Depending on what they are doing and just how they are doing it, neutralization might be followed with conductivity.
If I knew a little more about the process, I might be able to come up more definite answers.