You are on page 1of 5

DEPARTMENT OF CHEMICAL AND BIOMOLECULAR ENGINEERING

RESEARCH PROGRESS REPORT FORM

Name: Ajinkya Pal


Supervisor(s): Prof. Karimi I A , Dr Easa I Al-Musleh
Year in the Program: 2nd year
Dates of Applicable Time-Period: From: 15/08/2019 To: 15/02/2020
1. Oral Qualifying Exam

Passed OQE on 6th January 2020.

2. Summary of Research Progress:

Membrane based nitrogen removal process: Comparison of classic and superstructure


cascade design results

The cascade designs presented in literature largely divided into two categories: classic cascade and
superstructure cascades. Available membrane selectivity is limited and commercial applications
require recycling streams. Membrane cascade consists of multiple membrane stages requiring
compression and cooling before recycling the stream, as membrane selectivity properties are prone
to deteriorate at high temperature. The simplest form of the cascade, referred as classic cascade
design, consists multiple membrane stages with permeate from a stage after compression and
subsequent cooling sent as feed to the succeeding stage and retentate sent to the previous stage. A
generic superstructure cascade design, which allows more connection and deploys an optimization
strategy for identification of operating parameters, has drawn attention in recent years. The general
objective is to investigate and compare the number of stages and recycle compressors in stripping
and enriching stages of a classic (constant stage separation factor) and superstructure design based
cascade for binary gas separation to achieve a pre-determined value of recovery and purity with the
objective of minimizing the specific energy consumption. The study carried compares and assesses
the variation in specific energy requirement, the number of stages, total recycle flow and membrane
area to achieve a varying degree of purity (1,2,3,4 mol%) and a fixed recovery (93 mol%) make the
basis of four cases considered.

ProMax®5 software provided by Bryan Research & Engineering (BR&E), a powerful and versatile
process simulation software package was used to carry out membrane cascade simulation studies
for classic cascade design. ProMax®5 has an inbuilt membrane tool, capable of simulating
membrane as an asynchronous vapour phase separation process. Figure 1 compares the optimal
values of specific energy requirement and recycle flow for different cases and gives insight
regarding the energy penalty for increasing the purity with a constant recovery. The optimal values
for remaining cases were obtained by the similar procedure step as the first case. Each point plotted
on the graph has been obtained after 45 simulation runs. The purity and recoveries for each cases
exceeded the specified value. The specific energy requirement increased with increase in purity.

Figure 1: Optimal values for different cases.

Figure 2 represents the proposed superstructure and all the possible interconnection and splits
between the stages. Assuming ‘n’ stages in the superstructure, the main feed to the system is split
into ‘n’ streams and acts as the main feed to each of the n stages’ feed mixers. The separation
process is driven by the values of membrane area and the pressure ratio, divide the feed stream into
retentate and permeate streams, R and P respectively. The retentate stream is split into ‘n+1’ streams
with (n+1)th stream, referred as Rout, is the high-pressure retentate drawn from the sth stage of the
membrane cascade. The remaining ‘n’ streams act as retentate recycle stream, with each stream
added to the feed mixer of respective stages. The retentate self-recycle stream added to the feed
mixer; increase the recovery of the high permeable component without the expense of compression
energy as the retentate pressure drop assumed negligible. gPROMS ModelBuilder provides a robust
and powerful environment for creating and managing equation-based custom model with virtually any
level of complexity varying from a single unit to entire corporate process and simultaneously handle
high a number of equations and variables.

Figure 2: Representation of proposed superstructure


The case study was carried out for CASE 3 and CASE 4 to achieve ≥ 93 mole % methane recovery and
3mol% and 4mol% nitrogen purity respectively. The CSF design strategy suggests the optimal solution
with four stages for both CASE 3 and CASE 4. The model optimization results for two and three stages
for both the cases did not give any feasible solution. The model with four stages provided a feasible
solution and presented gave optimal solution similar to the CSF strategy. The decision variable of the
four-stage model comprised of pressure ratio, membrane area and splits of each stage. The values for
decision variables obtained after optimization were utilised to run ProMax®5 simulations, in order to
crosscheck the results. The values of specific energy and membrane area calculated in ProMax®5 for
super cascade approach resembled the optimal values calculated through our cascade model in
gPROMS, hence validating our membrane model against a simulation-based software membrane
module. The comparison is drawn between the results for CSF and superstructure design strategy
presented in Table 1. The specific power for the superstructure for both the cases was lower than the
CSF strategy, however, the margin was almost negligible. The compression power requirement
displayed a similar trend. The similar results for both approaches are in line with the conclusion
drawn in literature. The literature suggested that CSF strategy and variable stage separation factor
strategy provided a similar global optimal solution when the objective function was considered as
specific energy requirement. The high compression energy requirements owing to modest values of
selectivity and permeability even for modest value of recovery make membrane separation process
infeasible for practical application. The work gives us insight that for optimal specific energy
consumption, CSF and superstructure approaches give similar results and the former being a simple and
robust approach should be considered for designing of the membrane process, subjected to availability
of membrane with high permeability and selectivity.
Table 1: Comparison of optimal solution of CSF and superstructure cascade strategy.
Cryogenic based nitrogen separation process before NGL recovery

The cryogenic based NRU before NGL recovery, consisting of one high-pressure (HP) column, one
low-pressure column (LP) and two multi-stream heat exchanger(MSHE), based on suggestions
made in literature is delineated in Figure 3.The feed enters MSHE-1, cooled and fed to a phase
separator T-1, the vapour phase, cooled again in MSHE-1 and fed to phase separator T-3. The liquid
stream from T-1 and T-3, with pressure reduced to 56 bars act as cool stream and makeup product
stream P-2, P-3 and P-4. The vapour stream from T-3 and T-4, fed to the expander to reduce
pressure and make up the feed to HPC. The SPLIT-1 ensure a part of the top product from HPC
acts as feed to LPC and the other part compressed and cooled in MSHE-1, sums up the two reflux
stream for the columns. The bottom stream of HPC consisting mainly of methane, ethane and higher
hydrocarbon, makes up the NGL stream. The SPLIT-3 ensure a part of the bottom stream from LPC
acts as a reboiler stream to increase the purity of the product. The pressure of stream P-2/P-3, P-4,
P-1 and NGL is ensured to be ≥ 56 bars by valve V-T-1, V-T-3 and pump P-LB and P-HP
respectively. The streams P-1, P-2, P-3, P-4 and NGL form the HP product stream and act as the
feed to NGL unit, demanding operating pressure of 56 bars. The cold energy in the process,
provided by turbo-expansion, effected by valves and expander, is retrieved in MSHE-1 and MSHE-
2. The cold energy from the top product of LPC, the low-pressure residue is captured in MSHE-1.
The required minimum temperature approach (MTA) for the MSHE, the upper limit for nitrogen
content and target methane recovery for the process are 2, 0.5 mol% and 99.5 respectively. The
objective of the process is to minimize the specific power consumption.

Figure 3: Process flow diagram of cryogenic based NRU before NGL recovery
A simulation based optimization framework combining the accurate and fast thermodynamic
calculation performed in Aspen Hysys and rigorous optimization algorithm implemented on an
external platform like MATLAB, can provide accurate value for the decision variables listed in
Figure 4, while ensuring the constraints are met and specific power optimized. The connection
between MATLAB and HYSYS established through component object model in ActiveX, allows
direct two-way communication between HYSYS and MATLAB. The process and stream data
calculated in HYSYS are sent to MATLAB, which performs optimization, levies constraints on
variables, and sends back the values of calculated variables.

Figure 4: Optimization results.

Overall Research Progress: Satisfactory ______________, NOT Satisfactory ____________

Comments of the Progress Committee:

Representative of the Progress Committee: _________________________________________


Date: ________________________

You might also like