8/7/2019 Hydrocarbon Processing - Optimize heat transfer networks

1/7

H E A T T R A S F E R BONUSREPORT

O p tim ize heat transfer networksAn innovative m etho d ap plies h eat integ ratio n to co st-effe ctive lyretro fit b ottlen ecks in u tility system sM. THUBAITI, Saudi Aramco, Dhahran, Saudi Arabia; N. AL~AZRI and M. EL~HALWAGI,Texas A&M University, College Station, Texas

roper hear integration isessential for efficienc operation of anyprocessing facility. Significam research on heat exchanger net-work design has yielded rnu h needed improvements, New

developments combine debordenecking and heat integration efforts[Q improve total energy efficiency. This case study examines how roimprove an exisring exchanger network and process performancethrough a heat imegration approach.New approaches. A new debortlcnecking approach tharovercomes Iirnitarions of conventional sequential unit-by-unitdeborrlenecking approach has been discussed. IThis method issimultaneous in nature. It is based all posing [he deborcleneckingtask as a process integra ion task, which links all the design andoperating degrees of freedom and exploits synergies among theunits and streams [Q attain maximum debottlenecking.However, hear integration was nor considered in rhis approach.We will introduce a simultaneous approach co the debordenecking

and bear inregrarion, This method can be applied when retrofittingan existing hear exchanger network with a no/low cost strategy. Thepresented case study shows the applicability of this approach.Background. Proper hear integration is essential for the effi-cient and cost-effective operation of any bydrocarbon processingfacili~. Over the lase 30 years, significant research contributionshave been made in developing design techniques of heat exchangenetworks. Much of this work has focused on heat inregracion asthe main goal with supporting objectives such as minimizing hear-ing and cooling utilities and rotal annualized cost of the network.onversely, less work has been done on reconciling hear integra-don with other processing objectives.A key process objective is debottlenecking. For proflrable pro-

cesses with tighr capacity, there is an incentive to increase produc-tion output. As production increases, a processing unit or a processresource may reach maximum capacity and create a bottleneck, Insuch cases, ir is necessary to 'deborrleneck" the process in order toraise output. An important class of deborrle.necking is the no/low-cost approach in which no new equipment are added. It involvesmodifying the design and operating conditions and rerouting pro-cess streams, Such modilkadons may change unit heat duties.

Ince me focus of this discussion is no/low-cosr strategies, nonew hear exchangers, furnace, boilers or cooling/refrigerationsystems are [Q be added to the hear exchange network. Tills issueposes rwo challenges:

Maximlzing use of present utilities to avoid installing newboilers, furnaces or cooling/refrigeration systems

Optimizing effective utilization of existing heat exchangers[Q eliminate [he need to add new heat exchangers.To address these challenges, we propose to include a cornbina-

rion of heat inregradon and retroflrring of hear exchange networks(HENs) inro [he total design procedure when debotclenecking aprocess unit. Such retrofits consider the existing equipment andlayout and address the trade-ofls among energy savings, modifica-rion costs and deborclenecking benefits.

Retrofit an existing process unit and plant is complicated.Processing units are comprised of numerous pieces of equipment, anutility system and a heat-exchange network. These components arelnrerconnecred, modifying anyone of the components or equipmentitems within the system can and will affect pare, and possibly theentire sysrem. Retrofitting an hear exchange network may be con-sidered when trying ro reduce utility costs or as a result of changes inprocess streams or other operating conditions within me planr.L1 a typical petrochemical/chemical plant, hear integration is acritical element of the debortlenecking process. The difficulty ofincorporadng heat integration inro a debotrlenecking design lieswithin the srronginreraction between these two objectives. Oneway ro resolve this conflicr is [Qadopt a decomposition approach,where a certain extent of deborrlenecklng is related to a particularser of heating and cooling requirements. With the heating andcooling requirements temporarily fixed and all Flowrares andtemperarures of hot and cold streams are known, the minimumheating and cooling duties may be calculated.2-4 The procedure isrepeated and a tradeoff isestablished between the debottieneckingobjectives and heat integration. While this approach may be read-ily implemented and automated, it may be limited because it issequential. In addition, it may Fail to properly consider the srronginteraction between the process and potential hear integration.T his s eq ue ntia l a pp ro ac h L ea ds to s ub op tim a l s olu tio ns .Another approach developed a straregy for simultaneo LlS oprimiza-

don of the process and he-atintegration based 011mixed inreger Unearprogramming (MILP ) . ~ However, while rhe flowrares of the streamscan be [reared as continuous variables, the temperatures can only beassumed as discrete values. Others developed a model chat overcomesthis Iirnitarion.f The developers proposed a secof inequalities rhac relyon a pinch-location model and predict the minimum utility require-ments for variable flows and remperamres of the pmcess streams andfixed minimum temperature approach.fAsrnoorh approximation is used to handle the structural non-

differentiabili ies mat can arise in this formulation. This method isvery effective in handling a wide variety of hear integration prob-lems. However, care must be given to cases when m e approxirna-

H Y D R .O C A R BO N P R O C ES S IN G M A R C H 2 00 8 I 10 9

8/7/2019 Hydrocarbon Processing - Optimize heat transfer networks

2/7

BONUSREPORT H EA T T R AN SF E Rrion at some poinrs becomes ill-conditioned and for cases involv-ing errors associated with the heat loads of isothermal streamsand Intermediate urilitles. A superstructure representation thatincludes many possible flowsheer alternatives was introduced'?However, the Dumber of variables and constraints [hat are neededto produce the required mathematical representations can belarge. B Th LlS, sirnpi Hying assurn priam are reg uired.Another method for rhe simultaneous optimization of flow-sheet and heat integration was developed by Grossmann et al . .9 Itis

based on introducing integer variables that give a general formula-tion for bear loads and composite curves. This method overcomesme limirations of a smooth approximation method.

In a previous paper, a ne w approach for the simultaneous nollow COStdebortlenecking of a chemical plant was introduced.'Heat integration was not considered in this approach. We willintroduce a simultaneous approach [Q the debotdenecking andheat integration. This approach will consider the rerrofirring ofplanes heat exchange network applying a no/low-cost strategy. Acase study will demonstrate [he applicability of this method.Problem st.at.ement ..Assume a process with a certain feed-stock of raw materialts) and other processing units, which arereferred to as sinks. The set of sinks is SfNKS =0 [u:u '" 1, M s i n k . , 1and each sink has a set of input streams (INPUT,J and a set ofoutput streams (OUTPUT,,).The input stream, iu, has a flow rate G i " at temperature T i,,-

Each stream has a set K of desired components. The kth compo-nent has a composition referred to as xiu,"" Each sink has a rangeof acceptable fiowrareE and composition of species, and any screammust satisfy that range before being fed ro that sink, i.e.:

i E INPUT , U E SINKSII U (1 )i e INPUT, U E SINKS, k E Kr~ I~ (2)

TmiJJ. 5, T 5, T r r I D