OPTIMIZATION OF A REFINERY CRUDE DISTILLATION UNIT IN THE CONTEXT OF TOTAL ENERGY REQUIREMENT

1
OPTIMIZATION OF A REFINERY CRUDE DISTILLATION
UNIT IN THE CONTEXT OF TOTAL ENERGY REQUIREMENT

• E. O. Okeke A. A. Osakwe-Akofe
• NNPC RD Division, Port Harcourt, Nigeria
• APACT03, York, 28 30, April, 2003

2
INTRODUCTION

• The Nigerian National Petroleum Corporation, has
  4 refineries, in its downstream operations,
• The primary goal of this refiner is to achieve
  and maintain high gasoline production,
• Hence, the main objective of this study is to
  optimize gasoline production in all the
  refineries,
• The strategy being to first target the CDUs in
  these refineries. Maximizing the yield of
  gasoline and its intermediates will directly
  impact positively on total pool gasoline
  production,

3
PROGRAMME FOR MAXIMIZING GASOLINE PRODUCTION

• Maximizing gasoline and its intermediates
  production from the refinries has been planned to
  be accomplished in phases, viz-
• Phase I CDU 1 (the first refinerys CDU)
• Phase II CDU 2,3,4, 5 (the other 3 refineries),
• Phase III Catalytic plants - CRU, FCC HF Alky
• Phase I began with CDU 1 as a basis to ascertain
  plant suitability to process different crude oil.

4
CDU 1 FEED MAIN COLUMN SUBSYSTEM

• The CDU 1 of the first of these refineries, the
  object of our presentation, was installed in the
  1960s to process naphthenic crude of API 40.3 at
  first and another of API 35.4 afterward,
• It has a main fractionator with 44 trays and 4
  side strippers, and a stabilizer column.

5
CDU 1 DISTILLATES

• The intermediate distillates are as in
  conventional CDUs,
• Unstabilized gasoline from the main fractionator
  is further processed in the stabilizer column,
• Straight run naphtha and other distillates from
  the main fractionator are routed further
  downstream for processing and upgrading,
• Stabilizer produces an intermediate gasoline as
  bottoms and LPG as overhead

6
CDU 1 MAIN DESIGN HARDWARE FEATURES

• Licensed by SHELL and designed as a conventional
  crude distillation unit,
• Crude oil characteristics and product
  requirements as applicable in establishing
  hardware design,
• Hardware performance evaluation, maintenance and
  upgrading of facility undertaken periodically.

7
MAIN FOCUS AREAS TO ACHIEVE MAXIMUM GASOLINE IN
CDU 1

• Main areas are
• efficient operation of the CDU,
• review of configuration of CDU to determine
  opportunity for further increase in gasoline
  yield,

8
GENERALIZED STRUCTURE OF THE CDU 1

• The CDU can be decomposed in stages as follows
• Stage 1, the main fractionator producing feed for
  Stage 2 (i.e. the stabilizer)
• Achievement and sustenance of increase yield must
  be progressive from Stage 1 through Stage 2

9
METHODOLOGY STEADY STATE SIMULATION TO
OPTIMIZATION

• The main stages are as follows
• Compare the crude assays for the two naphthenic
  crudes,
• Configure, specification and steady state
  simulation of the CDU using HYSYS.Plant,
• Match HYSYS.Plant simulation results with
  original design requirements,
• Carry out optimization of the CDU
• Results obtained showed good opportunity.

10
COMPARISION OF THE TWO CRUDES
11
COMPARISION OF PRODUCTS DERIVED FROM ON THE TWO
CRUDES
12
INCREASING GASOLINE YIELD

• For a given CDU, yield of gasoline derivatives
  depends on,
• Feed characteristics,
• Process requirements/operating conditions.
• From the above therefore, since feed is
  constant, optimizing gasoline yield will depend
  on process requirements/operating conditions.

13
FRONT-END CDU 1 EVALUATION FOR HYSYS
IMPLEMENTATION

• The evaluation of the CDU is as follows
• Establish a reliable CDU configuration, determine
  process conditions using HYSYS and match these
  with the original plant design basis and
  requirements,
• Properly decompose the structure of the CDU and
  determine boundary conditions for optimization,
• Achieve a reliable process optimization in the
  context of total energy requirements.

14
OPTIMIZATION PARAMETERS

• The parameters for optimization are derived from
  process/hardware environments, viz,
• The main fractionator and the stabilizer are
  linked together stabilizer feed comes from the
  main fractionator,
• The other gasoline blending stock, SRN, a
  derivative from the main fractionator is routed
  for further processing,
• Four side strippers in the main fractionator,
• The stabilizer has a condenser and a reboiler

15
PLANT ARRANGEMENT FOR OPTIMIZATION
16
HEAT LOAD DISTRIBUTION

• CDU has an integrated heat exchanger network for
  heat recovery which shares loads, viz, Q1,,Q7,
  where Q4 and Q5 are utilities,
• Heat loads in the network are assumed to be
  efficiently shared,
• Heat supplied through the crude charge and for
  the various steam stripping supplies are
  constant.

17
HYSYS FLOWSHETET CONFIGURATION Overall CDU
18
HYSYS FLOWSHEET CONFIGURATION Main Column
Subsystem
19
MODELLING PROCEDURE

• Stage-wise approach was adopted, viz,
• Evaluate CDU configuration and steady state
  simulation data to determine opportunity for
  optimization,
• Based on the structure of CDU process and
  hardware requirements, evolve an optimization
  algorithm and define boundary conditions to be
  solved by HYSYS.Plant,
• Define steady state parameters from HYSYS.Plant
  simulation as first level data, and referenced as
  base or design values,
• Optimize the overall gasoline yield in the
  context of total energy requirement.

20
OPPORTUNITIES FOR OPTIMIZING GASOLINE YIELD

• We observed the following
• The columns are linked in sequential arrangement,
• Possibility of enhanced recoveries of gasoline
  in the nearest distillates below and above SRG,
  ie SRK and LPG, and in the stabilizer overhead,
• To maintain high quality gasoline to meet base or
  design specification, the path to solution must
  be constrained,
• Problem is non-linear.
• Based on these conditions an algorithm was
  developed

21
THE ALGORITHM Heat Loads

• Heat load differential at steady state
• ?Qibase Q1base Q2base Q7base 1
• Heat load at any level of optimization
• ?Qiopt Q1opt Q2opt Q7opt 2
• And the differential
• ?Qdifferential ?Qiopt - ?Qibase 3

22
THE ALGORITHM Gasoline Yields

• Gasoline yield at steady state
• ?yibase y1base y2base y7base 4
• Gasoline yield at any level of optimization
• ?yiopt y1opt y2opt y7opt 5
• And the differential
• ?ydifferential ?yiopt ?yibase 6

23
THE ALGORITHM Objective Function

• Incorporating the various energy and gasoline
  costs, the resultant differential becomes,
• INB ?ydifferential - ?Qdifferential 7
• The objective function becomes
• Max f(X1,X2,X3) ?ydifferential-Qdifferential
  8
• Where,
• ?ydifferential ?Qdifferential are gasoline
  and energy costs,
• X1, main column naphtha stripper reboiler return
  temp,
• X2, main column kero stripper reboiler return
  temp,
• X3, stabilizer reboiler return temp,
• Subject to RON and RVP of gasoline being within
  base or design values.

24
HYSYS OPTMIZER

• Primary variables (X1, X2, X3) are manipulated to
  maximize INB. Primary variables must have upper
  lower limits, and these are used to normalize the
  primary variables, viz,
• Xinorm (Xi Xilower)/(Xiupper Xilower).
  Where Xi X1, X2, X3
• Objective function as defined by INB,
• Constraints as defined for RON RVP,

25
OPTIMIZATION BY SEQUENTIAL QUADRATIC PROGRAMMING

• Sequential Quadratic Programming (SQP) was
  applied for solution.
• SQP minimizes a quadratic approximation of the
  Lagrangian function subject to linear
  approximations of the constraints. The second
  derivative matrix of the Lagrangian function is
  estimated automatically. A line search procedure
  utilizing the watchdog technique (Chamberlain
  Powel) is used.

26
PROBLEM SOLUTION

• Sequential quadratic programming was found to be
  ideal for solution,
• Solution was found for all cases studied,
• General increase in yield of stabilizer feed and
  SRN from the main column,
• Gasoline yield was increased by 8

27
BASE OPTIMIZED VALUES
28
TESTING ALGORITHM ROBUSTNESS RELATIONSHIP OF
KEY PARAMETERS

• Some optimization test runs were done using same
  HYSYS.Plant to
• Test the robustness and reliability of the
  algorithm at achieving early convergence,
• Determine the variation of key parameters, that
  impact on the structure of the CDU and the
  interaction of the main fractionator and the
  stabilizer. These parameters are the naphtha
  stripper reboiler return temp, the kero stripper
  reboiler return temp, and the stabilizer gasoline.

29
VARIATION OF GASOLINE WITH NAPHTHA STRIPPER
REBOILER RETURN TEMP
30
VARIATION OF GASOLINE WITH KERO STRIPPER REBOILER
RETRUN TEMP
31
VARIATION OF GASOLINE WITH STABILIZER REBOILER
RETRUN TEMP
32
OBSERVATIONS FROM THE OPTIMIZATION

• The optimization based on this algorithm achieves
  early convergence,
• As expected, the naphtha stripper (X1) and kero
  stripper reboiler (X2) temperatures have indirect
  impact on the stabilizer gasoline, while the
  stabilizer reboiler (X3) temperature has a direct
  impact on the same gasoline yield,
• The 3 parameters X1, X2 X3 are manipulated
  as appropriate to optimize the gasoline produced.

33
CONCLUSION

• Sequential quadratic programme technique ideal
  for solution,
• Solution of the algorithm is reliable, achieving
  early convergence in the cases studied,
• Objective of obtaining increased gasoline yield
  in the context of reduced energy requirement
  achieved,
• Since the configuration of the refinery CDUs are
  similar, this algorithm can be applied to
  optimize the CDU 2,3,4,5 in the other 3 refineries

34
(No Transcript)