ape SP a aes SB ee ee ee ee ee Troubleshooting Process Operations THIRD EDITION Norman P. Lieberman ' Pennwell Books PENNWELL PUBLISHING COMPANY | TULSA, OKLAHOMA,Copyright © 1991 by PennWell Publishing Company Dedication M21 South Sheridan Road P.O. Box 1260 To my por Tulsa, Oklahoma 74101 nis—Mary and Lou. Wisdom comes with experience in Iie. Library of Congress Cataloging in Publication Data Lieberman, Norman P. ‘Troubleshooting process operations / Norman P. Lioberman, — dre ed om, Includes bibliographical relerences and index. ISBN 0-87814.348.3, |. Petroleum retineries—Maintenance and repair—Handbooks, manuals, ete. 1 tte, TPS80.3.L53 1991 665.5'9'078—de20 90-28342, cp Al tights reserved. No part ofthis hook may be reproduced, stored in a retrieval sysiem, or transcribed in any form or by any means, electronic or ‘mechanical, including photocopying and recording, without the prior ‘written perraission of the publisher. Printed in the United States of America 2345 4 99Contents Dedication, v Acknowledgments, xi Proface to Third Edition, xili Prelace to Second Edition, xv Preface to First Edition, xvii Introduction, xix Section 1 SPECIFIC PROCESSES, 1 % 1 Crude Distillation, 3 2 Delayed Coking Cycle 3 Delayed Coking Process, 79 99 4 Amine Regeneration and Scrubbing. $0 5 Sulfur Recovery. 111 6 Alkylation, 196, 7 Fluid Catalytic Cracking Units, 186 @ FCCU Product Fractionation, 182 9 Saving Energy at Reduced Feed Rats Section 2_ PROCESS EQUIPMENT, 225 10 Refrigeration Systems, 227 11 Centritugat Pumps, 242 12 Distillation Towers, 256 18 Vacuum Towers. 281 14 Treating Liquid Hydrocarbons, 901 ais 16 Water Cooters, 999 17 Alarms and Trips. 280 15 Process HeatersCONTENTS tx ConmENTs Section 3 PRACTICAL PROBLEMS, 263 29 How to Make Field Measurements, S11 90 The People Problem, 522 18 Additional Distillation Problems, 965 19 Fouled Trays, 299 20 Dehydrating Light-Ends Towers, 298 21 Vapor-Liquid Separation, 406 22 Refinery Metallurgy for Novices, 41S 22 Unusual Noises and Vibrations, 490 Section 4 GAS DRYING & COMPRESSION, 499 24 Gas Drying. 441 456 28 Gas Compre: 28 Suggestions for New Process Operating Engineers, 479 27 Planning a Performance Test, 486 28 Understanding Control Board Instrumente, 498Acknowledgments The author wishes to thank his friends and colleagues in the major U.S. refineries who have supported his efforts to disseminate process tech. nology at their plants: + ARCO—Doug Arnold Chevron—Ken Rickter Exxon-—Richard Cotton Marathon—Eric Hennings Mobil—Dennis Huckaby | SOHIO—Paul Osterreich Sun—Greg Tradget Texacn-—-Ron Anderson Unocal—Rand SwensonPreface to third edition Ten years ago, | wrote the first edition to this book. For the past seven years, | have operated my refinery troubleshooting business. During this period | have worked through enough process operating problems and equipment design errors to fill a dozen volumes. Much of my business Continues to be conducted with the 15 or so major refiners in the United States and Canada, The tech service engineers employed by my clients, who work with me on these jobs, always ask what special techniques | employ 10 solve these problems. The procedure | use is the same one | cused 10 years ago: 1. Discuss the problem with the shift operators 2. Personally collect field data and carefully observe the operation of the unit 3. Develop a theory as to the cause of the malfunction, The error my clients often make is that they develop a theory, usually with process computer simulations, as to the cause of the malfunction, The theory is then reviewed with management and other technical personnel at a large meeting. If no one objects to the theory, it is accepted as the solution to the problem. Typically, no one at the meeting has discussed the problem or the solution with the shift operators, nor has anyone personally observed the process deficiency in the field, Finally, the intended solution is not put to a plant test to see if it is consistent with the problem. This approach to solving refinery process problems by the major oil companies often results in wasting capital resources and engineering man-hours I would be pleased fo discuss any problems or comments relating to the information imparted in this text as they may apply to specific operating, ‘or design problems. Norman P. Lieberman (904) 887-7714Preface to first edition Rummaging tough the atic, 1 happened upon my ol college text books, Mucho my dismay, thee eared volumes, elucidating the print ples of mas tans, aid Now, and diferental calculus, seemed tig incompretnsioleand ther nelevan. Aer 13 years of app the fonda metals of chemical enginering in 2 dozen refines, ft id ht Tel teody fo tht final exam in advanced thermexdynomies can, however, dgnose and rept any an al basement sump pump the accumulated deposits from the watt heater, | The years dedieated to operating designing, and troubleshooting pen: | lcum processes taught een kth Lom humid nights spe ster | ing the oa of giant steam turbines unde South Tes sky have schooled ie inthe challenges facing finery shit workers Such ae the experiences I with to relate inthis book, isnot only the facts, but the felings of working ina petioteum reinry that | hope to pas along Norman P. LiebermanIntroduction (One warm spring day, the telephone rang in my Chicago home, Once again, one of our refineries had a problem, Thus began a typical rouble: shooting assignment This refinery depended on twin plants to recover sulfur removed from the crude oil, The Environmental Protection Agency regulales emissions in refineries, and the ability to recover sulfur can and has limited refinery throughput On this day the refinery was in trouble. Due to a boiler-tube failure, brought on by a combination of bad luck and poor judgment, one of the {wo sulfur recovery plants had suddenly shut down, With only one plant ‘operable, refinery crude run had been reduced by 25%. Possibly as a politi- al yesture Wo corporate: headquarters —are likely for lack of anything better to do in a desperate situation—-t was called upon to help. To help! But to help do what? Was | supposed to advise on repair and start-up of the plant that was out of service? Was | supposed to investi gate the cause of the boiler failure? Maybe they wanted me to devise a method to squeeze more capacity out of the plant that was stil operating, The novice troubleshooter should realize at the start of his career that the | rman asking for assistance usually has only the vaguest idea what he wants done. More often than not, he needs help in diagnosing the problems and rnot with implementing the solution, By Saturday afternoon | had arrived at the plant site. Needless to say, most refinery failures occur on weekends. As | had suspected, based on prior experience, no data had been assembled for my review, no meetings had been set up to solicit my advice, and no instructions had been left to define my task. Actually, the professional staf had al forgotten I was coming and had gone home. ‘This was just as well. The troubleshooter should begin by talking to the unit shift operators. These people run the plant 24 hours a day and, although they don’t always know why something happened, they can often tell you what really transpiredxe INTRODUCTION ‘itera ritual exchange of pleasantries, | sat down with the shift foreman. We discussed the situation; rather, he talked and | concentrated on suppress- ing my impatience. There is something about being in a refinery on a Saturday afternoon that makes one want to get on with the job and get home. It soon became apparent that the refinery really was in a dificult spot, The remaining sulfur recovery plant was limited to 100 tons/day (1/0) of sulfur. The refinery normally made 130 T/D. Consequently, crude run had been cut by 40,000 barrels per day (B/D) to avoid emitting sulfur pollutants to the atmosphere. The daily penalty was huge, several times my total annual salary. Limited! As soon as any operator uses the word limited, the trouble- shooter should respond, “Which piece of process equipment is limiting plant capacity?” After some evasion, the foreman referred me to the chief ‘operator, who would be able “to answer my question in more detail.” (The psychology of dealing with chief operators is a subject unto itself, as ‘mentioned in Chapter 30. Suffice it to say that utmost diplomacy is always warranted.) The chief came right to the point. The sulfur recovery plant ‘was limited by frontend pressure. The hydrogen-sulfide feed gas (HS) would spill to the flare whenever the pressure in the feed drum exceeded 10 pounds per square inch gauge (psig). Figure 1-1 illustrates the setup. The control valve, upstream of the feed drum, was used to liold the pressure in the drum below 10 psig. [As the flow through the plant was raised by opening this valve, the pres- sure drop in the plant increased, As a result, pressure in the feed drum also ‘The plant superintendent had left instructions to avoid spilling hydrogen sulfide to the flare. The operators were merely conforming to these instruc- tions. Raising feed gas charge above the 100-T/D rate would cause the feed drum to exceed 10 psig, But why was the pressure control valve, which allowed feed gas to spill to the flare, set at 10 psig? The feed drum was designed to withstand much higher pressures. The chief operator informed me that they held to the 10-psig limit because "We always do it this way.” For a recital of the historical circumstances supporting this limit, 1 was referred to the senior shift operator—Mr. Leroy Jackson. (This name is fictional, but the story is tue.) Demonstrating the cooperativeness that refinery workers display when they really know what they are talking about, he explained the problem. “When we raise gas flow to the feed drum, its pressure goes up,” said Jackson. "This doesn’t hurt anything. Except at about 10 psig, the feed flow recorder reaches the end of the chart. Then we don't know the plant INTRODUCTION. xt cLoseD LINE To. PLANT A SUT OF SERVICE PRESSURE. TO FLARE CONTROLLER PLANT IN, Has }-—>) ‘SERVICE, FEED ‘oes coNTROL VALVE FIGURE 1-1 Pressure in the feed drum limited the plant's capacity feed rate, Of course, this doesn't really matter, but it's sort of convenient to be able to read this flow.” He was referring to the flow recorder shown in Figure 1. When 1 explained how important it was in terms of dollars per day to increase sulfur plant feed, Mr. Jackson registered surprise. “Why, if someone had told me what was up, I'm sure we could have done something to get that extra sulfur through the plant.” Well, we did do something; we installed a new range tube on the flow recorder. A range tube is part of an orifice plate flow recorder transmit- ter. The longer the range tube, the greater the flow that can be measured bya flow recordér. We doubled the length of the range tube. This increased the maximum feed gas flow that could be recorded by 44% (Le, the square root of two). Changing a range tube takes about 30 minutes and about as many dollars. ‘The next step was to increase the setting on the pressure control valve, 4xxii INTRODUCTION which spilled feed gas tothe flare, Mr, Jackson increased this sting from 10 pxig to 15 psig. The tsk involved climbing several ladders and turning 2 dial inside the local pressure controller box The hardest par in any troubleshooting jab always involves the human element. In this case, | had to convince the foreman to tey something new Itdoes very litle good to say, "There is now no rational reason for you not to increase the feed gas rate” The Wick isto allow the man making the ital deco onk ts is eo ones he cage ate ‘TROUBLESHOOTING PROCESS OPERATIONS Vth some apprehension the foreman issued instructions to the chel ‘s ‘operator to increase feed ever so slightly while making sure that the plant Wee Sid not slide Into oblivion a8 a consequence. By the end ofthe next dy, the refinery had reestablished normal crude runs; sulfur production had tisen to 125 TD. In retrospect, it may seem ludicrous to have cut back production be cavse ofa limited range on an unimportant lot recorder. Both the problem and solution seem to be the sort of thing local supervision should have handled. Perhaps troubleshooting ofthis type i too tsvial forthe trained chemical engineer ‘Mort refinery dificulties have a simple origin. However, this simple origin is clouded by false data, misconceptions, superficial observations, Snd thirhand reports W the answer was obvious, you woul not have ben called upon. Your technical training is one tool you take into the field to reveal the underline problem, but confining your investiation to technical areas will severely limit your chances of success. The capacity limit of this sulk plant was, in a sense, not due tothe small range of the flow recorder. A communications breakdown between the unit operators and first-line supervision had resulted in an artificial limitation. The troubleshooting engineer is most effective when he over- comes this failure to communicate, Ths is best done by personally gathering data, making direct field observations, and most importantly, slicing the opinions of the shift operators. This type of activity, when joined with sound technical training, makes a poweril combination with which to tackle tefinery problems ven the most competent and experienced operating superintendent can become ineffective when given an incomplete account of a problem He ie often too involved in administrative matters to find the time (0 £0 cout int the field and get the straight sory. This is 2 weakness common To all large organizations Inthe tale just rected, my contribution was to go to the most important source of data available: the shit operator. One should cut through al the layers of tpersision and ask the people who turn the valves, To summa size, the answers ae inthe fild—not inthe office | |SECTION 1 Specific Processes [ esarch for somthing once well known but long since forgottenCHAPTER 1 i Crude distillation 1 January 23, 1983, the Good Hope Refinery was shut down: ‘Our problem was simple: We had run out of money. AS the technical manager, ! was faced with a choice, | could become ‘unemployed, of | could become a consultant. | easily concluded that the latter choice would come across smoother when I told my mother about the latest development in my career. Now the first thing a consultant needs is clients; that was going to be | a problem because 1 did not have any. Fortunately, an ol schoo! chur, who was working for AXECO, heard Iwas in financial difficulties and offered to help. He suggested that AXECO would retain my services for one day to review the operation of ils crude distillation unit. Evidently, crude runs had fallen from 195,000 barrels per stream day (B/SD) to 192,000 B/SD ‘an the No. 2 crude unit, My colleague went on to explain that AXECO’s technical staff were lied up in planning reviews and therefore management hhad decided to hire an outside consultant to troubleshoot the problem. Figure 1-2 summarizes the process flowsheet of No, 2 crude unit, | began my investigation by asking the operators why they were limited to only 192,000 BISD of crude charge, They responded that the color of the furnace ol (see Fig. 1=2) was the limiting factor an crude charge. Whenever the crude rate was increased above 192,000 BYSD, they reported, he furnace oil color would take on an unacceplable brown tinge These observations surprised me, because if the furnace oil is dark, the Fluid Catalytic Cracking Unit (FCCU) feed cut (ce. the next lowest product draw) must be even darker. However, when | inquired as to the color of the FCCU feed drawof, the operators responded that this product ‘was right on spec. That is, it was black No, that is not right," | explained, “FCCU feed is not supposed to be black. Gas oil (the FCCU feed is often referred to as almospheric gas 3CRUDE DISTILLATION 5 ~ ese ft —eerenvy nani FIGURE 1-2 + Installetion error causes premature flood: Ing. = —rervawece ont 2) ez | or NE Live sesto FIGURE 1-1 + Typical crude unit cil) fed to an FCCU is supposed to be a translucent greenish-purple. Black gas oil means that there is crude oil residual components in FCCU feed.” Crude residual components (called resid) contain metals such as nickel and vanadium. These metals, especially the nickel, accumulate on the crack- ing plant zeolite catalyst. The nickel promotes hydrothermal reactions in the FCCU (or Cat"), Such reactions preferentially produce low-value fuel 5 as and catalytic coke, consequently reducing the production of more vali . able diesel oil and gasoline. Hence, black gas oil downgrades a refinery's PEMEATER ability to produce motor fuels. explained all ofthis tothe operators, but they disagreed. They explained that their FCCU feed was right on spec; it was black, it had always been black, and it was supposed to be black Does Anyone Know the Question? | now understood something important about No. 2 crude unit. The problem was not that throughput had slipped a few thousand B/SD. The8 TROUBL JOOTING PROCESS OPERATIONS real problem wih this distin unit was that i was not making 2 clean soit between te fst se cut (CCL fed) and he bots produc es And cern: have noted ths phenomenon only oo ten, That manage nen tends 1 concentae on thoughput rater thon on faetionation ‘Whenever one encounters fractionation problem, he fist pial to ,,r—C