You are on page 1of 4

LEGEND FC Flow controller Fl Flow indicator FV Flow control valve HGBV Hot gas-bypass valve.

It serves the dual hJnction of compressor protection and capacity adjustment during low-duty situations Compressor to be switched on next after Lead, and to shut down prior to Lead Compressor to be switched on first and to shut down last Function to skip prepro-

a
-,

Lag

er
Lood
~ Average ~cenariO tor all users load

Lead Override

grammed sequences PC Pressure controller PI Pressure indicator PlC PV Stondby Programmabie logic controller 380 kW (Two chilIers, each at 550;. of desjan dut y) 635 kW (Two chillers each ' I running at 90% of desion dutv) Pressure control valve Compressor to be switched on last and to shut down first. Normolly remains off indicator

Vivek Foster Phildon

Aggarwal Wheeler Stapleton

and S. Singh Eastern Pte. Ltd

Peok lood with one group of users, plus the rest ot overoQe lood
AI! users running at peak capacity

TC Temperoture controller TI Temperature XCV TV Temperature control valve On-off type valve Electrical or software signal

GlaxoSmithKline

900 kW (Three chillers each running at 87% of design dut y)

tion of equipment in parallel appears . pgrades of existing chiller facilities often turn out to be to be a simple affair, and doesn't too trickier than expected. often invite as much attention as it deAdding incremental capaci- serves. Actually, extreme caution ties may result in conflicts in control needs to be exercised for such capacity logic, eventually forcing prolonged enhancements. Expansions desired in shutdowns to straighten out the prob- a hurry onsite are generally comprolem. The root cause usually traces mises -not the way one would expect back to a rushed modification. Some- new plants to be. one has made a bad choice to build imOur aim in this article is to illustrate mediatelyand analyze later. This artithe pitfalls using an ostensibly simple cle provides and example of a capacity problem. The objective is to upgrade an expansion of a glycol chiller system, existing chiller system's capacity by one-third. Glycol chillers provide the and shows how adding a third chiller cooling to diverse batch users. Due to to a set of two could cause instability. Then it describes how to control the the batch production, the number and duty of customers fluctuate considersystem better . Pharmaceutical facilities undergo ably, causing cooling demand to vary substantially all the time. periodic expansion and diversification programs, mostly toincrease manufacsetup turing capacities. Production is in Pre-expansion Before expansion, the chiller system batch processes,where several similar or dissimilar batches may be handled comprises two glycol-chiller packages, in the same equipment. Equipment normally operating in parallel (Figure has to be amply flexible to facilitate a 1). The warm-glycol temperature serves as input to a PLC (programmavariety of processing schemes -including some that were not even envis- ble-logic controller, dedicated to the aged during the inception stage. Such chiller compressor) to turn the chillers batch operating conditions with chang- on or off under various situations. The ing demand patterns require e:qually PLC decides if one or both chillers should be online, and has a proresponsive support utility facilities. grammed imposition ofmanufacturer's I Expansion of such manufacturing criteria: minimum run times, units typically occurs by increasing the safety trip scenarios and stoppingsequences. number oftrains operating in parallel. The warm and cold tanks are balSimilar increases in support-systems capacities through proportionate addi- anced, to allow passage of cold glycol

into the warm glycol tank whenever process usage decreases. This configuration holds the supply temperature to users constant. Even better, the flow through the chillers is also being held constant. Upraderequirements Due to the plant expansion, the total cooling demand will increase. The expected usage pattem will become as shown in Table 1. This calls for installation of one additional chiller to cater to the simultaneous peak demand of all the users. The "quick" proposal One meeting yields a "quick" proposal. It looks like expansion should be easy: just add a third chiller in parallel to the existing two chillers (Figure 2), and add a third 100-m3/h warm-glycol pump. Also, the plant must tap a new header from the cold-pumps discharge manifold to run to the new user facility. Engineering's task is to implement this idea. The users of the system will be needing an increased flow of cold glycol. Clearly, the system requires additional pumping capacities to cater
to the an increased chillers, due warm-glycol to an flow additional to

chiller in parallel. The warm-glycol flow to chillers needs to be increased by 50 m3/h, so

CHEMICAL ENGINEERIN

we have a little margin. The additional pump of 100 m3/h in parallel changes the total pumping capacity to 200 m3/h. Rut now all pumps will operate at below rated flow, thereby causing power waste. Rather, replace the existing warm pumps with new, higher capacity, 150-m3/h pumps. The cold-glycol pumps are not controlled. Their flow fluctuates depending upon the number of users online. Glycol flow ta online users is modulated by valves to control the process temperature. So, it is all right ta add another cold-glycol pump similar ta the existing ones, so that two run in parallel with one in standby mode. Here, higher pumping capacity than normally required is actually desirable, since it enables these pumps to deliver a higher head at required lower flow. Design review of controls Looking into the finer details of the "quick" proposal raises a multitude of control questions: 1. How will the new chiller be controlled alongside existing chillers? 2. Will the existing PLC control it? 3. Does the existing PLC have any spare capacity? 4. How many chillers should normally run? 5. How will the upgraded system adjust to varying loads? 6. Can the new chiller integrate with the functioning of existing chillers? 7. Are the existing and new compressor-control software compatible? (A compressor is the heart of a chiller .) 8. Can the existing chiller's control system allow interference from a third chiller? 9. For that matter, can existing chillers function as standalone units? 10. As pumping capacities in the cold and warm glycolloopsare designeddifferently, is there a causefor concern? 11. What is the impact on supply temperature and flow to existing users? 12. Is the increased volumetric holdup among users a cause of concern? 13. Have the existing warm and cold storage tanks sufficient capacities? Analyze the upgrade We have written exact details of the control logic and operation for the chillers in the box, "Chiller Control Phi86

FIGURE 1. Cold glycolleaves the chillers under temperature control and collects in the cold tank. From there it is supplied to the users through cold pumps. The cold-glycol temperature controlIer varies the respective chiller's compressor capacity, through adjustment of the compressor's slide valve. Process users use glycol to control process temperature by adjusting the control valve on the glycol return. The warm glycol returns to a warm tank and is pumped back to chillers, thus completing the loop. FIGURE 2. The "quick" system expansion added a new chiller with TC and FC, one coldglycol pump, and associated piping. The warm-glycol pumps were reDlaced with a larger size. But, because of problems with unbalanced pumpIng and controllogjcs, it was necessary to replace the ~ obsolete PLC and do some control revisions (depicted inside the yellow rectangle)

.@":l

lr
--eJ
Ilvcol pumps ~ I = .1 = ~~~, per ~:i~lingicif/~

---@
New m3/h, 150 Dr---" Warm#Ji-.,ia.i 150m3/h, New standby
oplion-1 I

New 105 m3/h, operati"g Gold-glycol pumps

losophy," on p. 87, for readers who like details. Reading it helps you to understand operation of the loop while fully analyzing the upgrade. The existing chillers were identified to have several control constraints: .The existing chillers get switched on and off in response to glycol- return temperature via a common PLC .Existing compressor-control panels communicate with the common PLC for capacity determination, and so determine whether one or both existing chillers should run. Based on computed operating capacity, this PLC controls the operation of a hotgas-bypass valve (HGBV), and also has a part to play in operation of evaporator expansion valve. The HGBV also switches on for a predetermined time every time the compressor switches on .Existing chillers had several trip scenarios for machine safety; some activated via the PLC, and others through individual compressorcontrol panels

.The existing common PLC is obsolete. Newer models of PLCs do not allow a download from this old model, but can be reprogrammed with the same logic .There is some spare capacityin the existing PLC, but apparently only enough to just do last on, first off control of the new compressor .The new chiller control system's software is different from that of existing chillers. So, the new chiller compressor's control panel may not be able to commun~cate directly with the existing PLC. However, some new PLC models can communicate directly with the new and existing compressor control panels. Back to the drawing board It thus became clear that more engineering work was required. The objective became to minimize change in the existing setup and achieve the desired functionality. We came up with five options as listed in a box, p. 88. We

CHEMICAL ENGINEERING WWW.CHE.COM 2002 JUNE

CHILLER CONTROL PHILOSOPHY


E ach glycol chiller package has its own refrigeration system capabie of cooling glycol down to -20C (Figure 3). Refrigerant liquid enters the evaparator colder than -20C, and by exchange with the glycol exits as a superheated gas. Controller C -1 ensures superheat by reading the temperature and pressure at the vapor outlet and adjusting the expansion valve V-1 accordingly. Superheat is necessary because the compressor must not have any liquid in its feed. The other side of the evaparator converts warm glycol into cold glycol. Compressor K-1 raises the pressure and temperature of the refrigerant. lts warm gas is condensed in the condenser by heat exchange with water. The bulk of the condensed refrigerant goes to the economizer and then to expansion valve V-1 that in turn cools it down below -20C Two small streams provide fine control. A smallline of refrigerant is led through the economizer, where its vaporization cools the bulk of the refrigerant. This stream then enters into the compressor's sideport, thereby serving to increase compressor efficiency. Valve V-2 maintains a constant superheat byadjusting the flowrate of vaporized refrigerant, depending upon the compressor's operating capacity. In a screw compressor, operating capacity is determined by adjusting the position of a slide valve (not shown), which is varied to control the temperature of the cold

+ @

TI!

I
glycol
V-1 ..I~ ~ ~

! j i , , ,
: I!

l ,
.

.: ~t i -HGBV,

V-2 "" ~y~2~

~ C1 t

Condenser

FIGURE 3. Operation of the compressor shown here is explained in the text above glycol. The hot-gas-bypass line for the condenser through the HGBV is activated to protect the compressor, and adiust capacity when in a low-duty situation. D

FIVE WORKING Option 1: Discord the existing PLC. Provide 0 new PLC designed for communicoting directly with 011the chiller control ponels. Duplicote 011 possible necessory controls ond safety trips from existing PLC into the new PLC. Ensure thot the new PLC octuolly communicotes os designed. This option requires 0 lot of work, but is 0 good longterm solution. See CEEXTRA at www .che.com for 0 sample operoting sequence corresponding to this option. Option 2: Retoin the existing PLC. Provide on odditionol new PLC for communicoting with the new chiller's control poneland the existing PLC. The new PLC would not communicote with existing chillers

OPTIONS

direclly. This option ensures minimum impact on existing chillers, minimizing risk and downtime. But more space is required. The very old existing PLCwill remain as the weak link in the system, probably forcing review after more years of operation. Option 3: ( 1) Ensure that ex isting chillers can operate standalone at normal duty (without the existing PLC). Chillers are already controlled by cold glycol temperature at respective chiller oullet. (2) Discard existing PLC.Provide a replacement PLCcapable of doing the following : (a) Switching on the chillers, based on the cold glycol temperature. Whenever 'set' temperature of
terms 1 sugthird cothe

-20C is exceeded for a preset time, switchon the subsequent chiller. (b) Switching ol! one chiller whenever operating capacity of chillers approach a "minimum." This minimum should be set such that opening of the HGBVs is totally prevented, unless dictated by individual compressor control panels. Option 4: (I) New chiller has its own separate controls. It always runs as either the lead or the standby. It cannot run as a lag item. In standby mode, it's on-ol! switching is controlled by TC-03, based on on warm glycol. (2) Two existing chillers continue to be controlled by the existing PLC (3) Set points in existing PLC

are adjusted, such that if one of them starts behaving like lag, the other becomes standby. Setpoints are adjusted to prevent the existing chiller's HGBV trom opening, unless dictated by individual machine's tripscenario tor machine protection (and this is controlled by the existing PLC). (4) Control the chillers using cold glycol temperature at respective chiller outlet. Option 5: Operate only existing chillers as lead or lag, as in existing plant. The new chiller will be standby and switched on and off by TIC03 during extremes of glycol temperature. No new PLC is required, but operational flexibility is not achieved.

went

with

Option

1. Unfamiliar p.85. in Table of in

are listed gests chiller incident units. mally result, normally tial load, standby. more (that that will This

in the Legend, pattem the peak won't

The loading

operation demands chillers the third fast of

be required

only during all

be a frequent

occuras a to on with chillers flow at 50 The incon-

ance, so all three operatingat operate with This uniform is, without evaporator flow

are not norcontrols chiller

the same time. two the chillers

we contigured

on parresponse

provides operation frequent that

to increased

dut y requirements, on-offs). the glycol

We can assume to each dividual 88 m3/h throughout

is constant its operation. provide

controllers

trol. An XCV on the supply line will switch off the glycol supply to the offline chiller . The temperature of the cold glycol exiting the evaporator is also controlled using TC-01/02/NEW at -20C. TC-01/02/NEW adjusts the respective screw compressor's operating capacity. Hence, the compressor capacity would modulate in response to variations in the return of hot glycol temperature to the chillers, resulting from changing the user's load. The online chillers will thus automatically and equally modulate their capacities, to respond to changing user requirements. TIC-03 (temperature indicator-controller on warm-glycol feed to chillers) tums the standby chiller on or off. The JUNE 2002

standby ehiller starts up only when the return temperature (ofTIC-O3) increases beyond -12C despite two chillers running at funload. Each chiner is fitted with its own HGBV. This valve anows the refrigeration systems to be run on zero load while keeping the compressor running in a running mode, and not tripping on low-pressure cutout. The hot-gas bypass arrangement win be used only as a protection for individual chiners. A HGBV win not be relied upon to deliver low process duties. Under normal conditions, a HGBV can only open when its system (chiner) is selected as the lead system; in other words, more than one HGBV cannot open at the same time. An exception to this rule is the unlikely event of two or

)M CHEMICAL ENGINEERING WWW.CHE.CC

more systems running with their 20min overrides activated and the heat load reduced. In such a case, it is possible for two or alJ three HGBVs to be on together. We propose to modulate chiller capacities from 100% of design capacity to as low as possible, to increase the range of chiller operation and to reduce the frequency of chiller cycles. Every time a chiller switches off and back on, there is a flow through the bypass line from a warm tank to a cold tank. This impacts the supply temperature that the users receive. By reducing frequent on-offs, we assure a more-uniform temperature suppJy to the users. Capacity set-points should be adjustable on-site.

Dissimilar pumping capacities The pumps differ in size and, since the balancing line between the tanks must always be open, we found an additional problem. The return flowrate of glycol to the warm tank must be less than the total feedrate to chiller units. holdup Otherwise we see glycol flowfrom the I lnereased volumetrie warm to the cold tanks. Hut, since The upgraded system has substanwarm-glycol pumps are bigger, warm tially more glycol inventory .Holdup glycol will pass over from the warm to capacity of the existing glycol tanks the cold tank most of the time. As a may become limiting. During shutconsequence, the temperature of the down, the glycol inventory tends to setchilled glycol to all users creeps higher tIe down to an equilibrium level. Glythan the design value of -20C, and col in the piping and equipment drops that rising temperature will gradually into the glycol starage tanks. We entend to become even warmer until the sured adequate storage for shutdown. next chiller comes online. This will result in an erratic temperature supply Discarded pumping options to the users. We had a few options to solve the We discovered this complication by pump imbalance problem. For compleconstruding the complete glycol circuit ness we summarize the unused ones. as a spreadsheet and running dynamic 1. Buy a warm pump with capacity simulations. We ran a series of heat of 210 m3/h, to operate the chillers at and material balances.Table 2 provides 70 m3/h instead of 50 m3/11. This would a sample of computed supply tempera- ensure that the cold- and warm-glycol tures. We confirmed that the tempera- systems were ba1anced as in existing ture to users win gradually become setup. However it would cause wideeven hotter as the cold-glycol supply spread change in chi1lers, FCVs, piptemperature to users becomeshotter . ing and instrumentation. Pressure drop through the chi1lers would increase, LlT would drop, heat transfer What caused this? The problem was caused by one or coefficient would increase, and LMTD would increase. Though the work inmore of the following: .Changing flow betweenthe warm and volved would be expensive, solution cold tanks (via the balancing line) reached would offer stable and pre.Change in nuffiber of online chillers, dictable operation. We did not have as cold- and warm-glycolloops will the time to implement this. I have different flowrates 2. Control the warm-glycol tempera-

1 ture returning to the warm tank at .Change in number of customers .Change in user flow requirements. -12C. Providing a temperature-control The customers circulate glycol for tem- valve between the supply and return perature control. So, increased supply could control this temperature return. temperature results in more glycol flow, That would mean that the bypass temprobably causing the distribution perature were always steady at -12C. pumps to operate lower on their curves As long as bypass flow remained and thus function less efficiently. More steady, supply temperature to users warm flow will cause a cascading effect would remain constant, although at by further increasing the cold-glycol warmer than the design value of -20C. temperature, aggravating the situation. However, this solution doesn't offer total stability as it stillleads to disturFix the disparity bance (undesirable unpredictable flucIt is fortuitous that a new PLC is part of I tuation) whenever bypass flow the upgrade becausemore controls are changes. For example, a change in the needed to alleviate the problem. The number of online chillers would preferred solution is to add a flow con- change the bypass flow. Changed bytroller to force the discharge rate ofthe pass flow at -12C would lead to cold pumps and the flowrate of the change in the supply temperature to warm pumps to match (Figure 2). Then, users. Similarly, every time there bypass will always be zero. Addition- were a change in the number of onally, provide a pressure-controlled flow line users, the bypass flow would also bypass for the cold pump, so that excess change -leading to a change in supcold glycol can be returned in case cus- ply temperature to the users. . tomers (under individual temperature Edited by Peter M. Silverberg control) are using lesser flow. Authors
Vivek Aggarwal is the chief process engineer for Foster Wheeler Eastern Pte Ltd (32 Maxwell &ad #02-03, Singapare 069115; Phone: 6568900671; Fax: 65-63242890; Email: aggarwalvk@hotmail .com). Prior to this he worked for the processdivision of Engineers India Limited. Ris work comprises conceptual studies, front-end engineering design, revamp, capacity upgrades and debottlenecking. He has ten years of experiencein basic processdesign and engineering of refineries, pharmaceutical, crude oil and natura!-gas facilities. He has written an article on the recovery of natura! gas liquids .He has an honors degreein chemica! engineering from Banaras Hindu University , India and two graduate coursesin finance and operations management. Satendra Singh is director of engineering, process and technologies with Foster Wheeler Pte Ltd(Phone: 65-68900671 Fax: 65-63242890; Email: ssingh@hotmail.com). is reHe spansible for process design and engineering functions. Earlier experience was with Reliance Industries Limited and Engineers India Limited. He has a B.S. and M.S. in chemica! engineering from the University of Roorkee. Pbildon R Stapleton is a senior project manager with GMS Engineering, Technology & Capital Management (1 Pioneer Sector 1, Jurong,Singapare 628413; Phone: 6568608-310; Email: prs84501@gsk.com).He is in charge of active-inl(redient pharmaceutical prooucts and support facilities. Prior to joining GSK, he was with Bush Boake & Allen Ltd. He graduated from the University of Surrey with an honours degree in mechanica! engineering and is a Chartered Engineer (UK) and a member of the Institute of Mechanical Engineering.

90

CHEMICAL

ENGINEERING

WWW.CHE.COM

JUNE 2002

You might also like