water is a very non-ideal fluid and the cp values are not constant even in small ranges of t, p,
you should define a maximum acceptable error for yor model,
simplified correlations (based on one or two parameters) can produce relatively large errors (i.e. 5-20% or more when applied outside a relatively limited range of t,p), if you need to follow that way (for example because the limits in DCS programming do not allow to include complex models) then you could break down the whole range and include several correlations for limited t,p ranges, this usually works, Wagner and Pruss and others have described many different simplified and complex models which you may adopt.
A well tested solution is IAPWS formulation for industrial use (IAPWS 1997) as suggested by Snorgy,
you can easily find the source code on the net,
finally if you need the best accuracy you may adopt the IAPWS formulation for general and scientific use (IAPWS 1995),
it is a quite complex model to program,
a solution could be to link an external dll (if your DCS allows that) which provides all the calc's,
this can be useful if you need to do some simulation as for example turbine, flash operations etc.
I use Prode Properties (which includes IAPWS 1995), other well known brands could be Chemicalogic and others.