I've completed (sort of) the computation of the process flowsheet for the isooctane process. The figure below shows the process flow diagram.
The new operations since the last post:
Pressure letdown valve - lowers pressure for the distillation column to reduce vessel cost.
Distillation column - 18 ideal stages were required. The DCOL routine would not solve with a total condenser, so a partial condenser was specified. It appears that the condensing temperature (193 F) is too different from the top temperature of the column (237 F) for DCOL to solve.
Condenser for column overhead - The overhead is primarily recycle solvent, so I wanted to use a pump instead of a compressor. Thus, another condenser was added.
Pump - for recycle solvent.
Compressor - for recycle and makeup hydrogen.
Preheater - for liquids to the CSTR.
Gas cooler - for hydrogen to CSTR.
I used direct substitution to converge the recycle gas and liquid streams to the CSTR. The outer loop converges in about 4 passes. The hydrogen and solvent makeup were computed based on consumption and losses, but the pentenes makeup stream was fixed. The pentene makeup rate establishes the production rate.
The PFD program
The drawing above was made with the free online program available from the Engineering Toolbox website http://www.engineeringtoolbox.com/pfd-process-flow-diagram-online-drawing-d_1640.html .
The program includes a drawing legend, but I removed it for this blog.
"Sort of" complete?
The flowsheet streams are converged so the computations are complete in that sense. However, the design may not be complete because economics have not yet been determined. Some design variables, such as the column pressure, might need changing to optimize the process.
Obtaining a nameplate capacity
Most process simulation programs can scale all streams to achieve a desired production rate. You should use that feature with caution. In some cases, a unit may have used parameters that depend upon rate. The performance of that unit will change with the flow rate. Thus, a direct scaleup would be inaccurate. As an example, in this process, the fixed bed pressure drop depends upon the flow rate. This would probably not matter for this example because of the pressure letdown valve immediately after the reactor. A better example would be a tubular reactor with heat exchange. For that example, the process should be re-computed until the production rate is met. As a general rule, realistic reactor models will have rate dependent parameters.
A working simulator, but not perfect
There are still improvements to be made, but I have demonstrated that the Prode Properties/Mathcad combination can be used to build a flexible, basic process simulation program. I want to explore some options for unit reports which are currently missing. However, the blog will move on to a new subject with the next post.
Next: tracer studies for fluidized beds