Foam control methods in delayed cokers
Transcription
Foam control methods in delayed cokers
REFINING Injection of high viscosity PDMS fluids for controlling the buildup of foam in delayed cokers reduces costs while at the same time increasing capacity and maintaining downstream quality control LawrenceN Kremer BakerPetrolite Timothy G Hueston Dow Corning have been used for decades to control foam in refinery delayed coking units. such as asphaltenes and resins, Since PDMSacts as a defoamer (to knock are a prevalent cause of foaming in down existing foam) and as an antifoam delayed cokers. If foam is allowed to (to prevent the build up of new foam) in build to a point where it is carried over the coke drum, both terms can be with the vapour takeoff, it can cause applied to the silicone oil used in the severeand costly downstream fouling or coking process. For simplicity, we will refer to the PDMS as a defoamer. even necessitatea shut down to unplug Over the years, it has been observed the vapour line. Therefore, it is critical to keep this carryover from occurring. in the field that increasing PDMSmolecular weight (as indicated by higher visTo prevent carryover, refiners will limit the level to which foam is allowed to cosities) results in lower dose rates; build in the coke drum. Once this level Initially 12500 centistoke (cSt) fluid was is reached, feed will be switched out of the viscosity of choice, but that has the drum to prevent further level gradually evolved to where 60000cSt is now more commonly used. In the past increase. There are typically two ways to reduce few years, there has been a movement the overall foam height. The first is to toward even higher viscosities, such as change the coker operation to give the 100 OOOcStand 600 OOOcSt,and even foam head time to break before it reach- 1 million cSt is now being considered. es the top of the drum by either lower- All other things being equal, such as ing the feed rate or lowering the target adequate material handling capabilities, capacity for the drum. Needless to say, higher viscosity PDMS appears to be these operational adjustments can rep- more efficient for delayed coking. A negative aspect of the use of siliresent a significant economic penalty. The second method is to use a chemical cone defoamers in the coker is defoamer or antifoam. to control the hydrotreater catalyst poisoning by silicon species. The coking process profoam head. Given that this second method allows duces products high in unsaturation greater capital utilisation, especially for and sulphur. For this reason, it is comexisting assets,and is usually quite inex- mon practice to hydrotreat coker naphtha and coker gasoil. It has been pensive from a material standpoint, chemical foam control is the most pop- observed in many refineries that the hydrotreater catalyst is slowly poisoned ular method of the two. Standard foam control theory says by silicon containing species, causing that a good foam control additive will be the activity to decline until the expenmore surfaceactive than the foam stabil- sive catalyst must be replaced. The ising surfactants present in the system, source of this silicon contamination has be chemically inert in th~ system,be dis- always been assumedto be the defoamer persible, but not soluble, and of course, used in the coke drum. There are methods to remove the silibe cost effective as compared to both con-containing species from the mechanical and chemical alternatives. Some silicon-based materials, particu- hydrotreater feed stream. Perhaps the larly polydimethylsiloxanes [Me3SiO- easiestof these is to run the hydrotreater (MezSiO)x-SiMe3, where x = 0 to >5000], feed through a guard bed containing a are very effective foam control additives relatively inexpensive sacrificial catalyst (or a wide range of chemical processes to adsorb/react the silicon speciesbefore they reach the actual hydrotreater cata[Clarson S] and Semlyn] A, Siloxane Polymers; lyst. The sacrificial catalyst can then be Prentice Hall Polymer Science and Technolochanged out periodically. However, the gy series, 1993]. Polydimethylsiloxane (PDMS) fluids labour and expense of this method usuV apour presence generation of natural and the surfactants, 65 PTQ SUMMER 2002 www.eptq.com ally makes it less attractive than just dealing with the fouling of the hydro treater catalyst. Since physical removal of the silicon speciesfrom the coker products is usually not attractive, there have been many attempts to eliminate or reduce the silicone defoamer usage in the coke drum. While chemical alternatives to PDMS have not been very successful,there are methods to minimise the defoamer consumption, including' proper addition methods and using the higher molecular weight PDMS as the active ingredient of the coker defoamer. Addition methods Several addition strategies have been tried for the defoamers to minimise the use of PDMS, maintain good control of foam, and produce the desired coke product. Severalpoints of addition have been used, including into the hydrocarbon feed to the drum, through nozzles on the side of the drum, and through nozzles at the top of the drum. Most refineries have now settled on using nozzles at the top of the drum. Maintenance issues are an important factor for injection nozzle location and design. While side of the drum injection nozzles can efficiently spray the defoamer on the surface of the foam head, they tend frequently to become plugged with coke. Since top of the drum nozzles are least likely to become plugged in this way, they are the most common system. Sophisticated nozzles have been tried, but they can easily be fouled or plugged in the coking environment. Many refineries have settled on injectors that are straight pipe cut 90° at the end. Injection quills that extend into the drum have been rejected since they can be bent during the cutting process. The purpose of the injection system is to get the PDMS molecules to the foam front as efficiently as possible, so that no defoamer is wasted. Defoamer that is entrained into the vapours leaving the drum will never reach the foam front to affect the foam in the drum, REFINING but be carried over to the fractionator. Likewise, defoamer that is on the walls of the drum will not efficiently defoam. Therefore, the injector is typically located as close to 1800 as possible from the overhead vapour line to minimise entrainment of defoamer in the overhead gasses. The injector should also be at least O.S-lm _(1.S-3ft) from the vertical walls of the drum to prevent the defoamer from running down the wall of the drum. Figure 1 depicts two typical injection systems. The defoamerspurchased by refineries are typically dilutions of the PDMS in hydrocarbon solvent. Even the 60 OOOcSt silicone oil is so viscous that it can only be pumped using expensive specialised handling equipment. The defoamers basedon diluted PDMShave much lower viscosities «1000cSt.), so they can be easily handled using standard equipment available in the refinery. Many refiners have found it helpful to further dilute the as-receiveddefoamer to obtain optimum performance. . Dilution can help in severalways.For example, increased flow through the injection quill increases the velocity, which better sprays the defoamer onto the foam front. Also, dilution of the defoamer enables the molecules to better spread across the foam front. As an added benefit, the coker naphtha and coker gasoil commonly used as diluents appear to provide some additional knock down capability in some cokers. While coker naphtha is sometimes used as the diluent, coker diesel or coker gasoil are often preferred because the higher boiling range solvents are less likely to flash off when injected into the hot coke drum. Two basic types of.defoamer dilution systemsexist. Many of the early designs employed a day tank system as illustrated on the left side of Figure 1. The defoamer and diluent are added to the day tank. The tank is mixed either with an impeller or air sparge.A 10:1 diluent to defoamer ratio is common. The diluted contents are then pumped to the injection quill at the top of the drum by means of a metering pump. In some cases the metering pump can be varied manually or electronically through the control system. Other systems rely on a small pump and a big pump in parallel. In any case, there is usually some flexibility in the pump system for the operators to adjust the injection rate of the diluted defoamer for variations in the foaming conditions. While this system is straightforward, there are many disadvantages.Since the make down system is typically a manual batch process, it requires significant In line Day tank Figure 1 Injection systemsused to meter defoamer into coke drums: automated inline injectionof defoamergivesmoreflexibilitythan day tank system operator involvement. Some systems require the operator to open the day tank and pour in defoamer, thus exposing the operator to hydrocarbon fumes. It is common for the dilution ratio to vary from batch to batch. Once the day tank has been filled, the only way to change the amount of defoamer injected into the coke drum is to change the metering pump settings. If the operator is distracted by other duties, it is possible for the tank to run dry. For these reasons, many refineries have designed and installed automated inline injection systems. The inline system is shown at the right of the figure. There is a continuous slipstream of diluent, typically coker gasoil. The defoamer product is metered into this slipstream with a variable injection pump. In this way the dilution ratio can be easily changed. It is also possible to easily increasethe amounfof defoamer injected into the drum. This automated system also frees the operators for other duties. Refineries have optimised the defoamer injection using the systems previously described. In order for refineries to further reduce silicone carryover it is now necessary to optimise the defoamer product as much as possible. The rest of this discussion describes how the viscosity (or molecular weight) of the silicone oil affects defoamer usage and silicone carryover. Theory The theoretical explanation why bigger silicone molecules can be better begins with the recognition that the operating temperatures of delayed cokers (800+oF) is well above the initial thermal decomposition temperature of PDMS (about 600°F).Fortunately, the polymers do not just {'blow apart" once this temperature is reached, but slo"'ly degrade at a rate dependent on temperature. There are several different thermal degradation mechanisms in the absence of oxygen, primarily molecular bond interchange and chain-end cyclisation and cleavage.Work over the past decade points to the chain ends as being very important ~o siloxane polymer stability, so it is thought that the chain end cyclisation and cleavagethermal degradation mechanism will be dominant in this application. In this mechanism, the chain end curls back on the rest of the polymer forming a small loop. The bonds then rearrange to form a cyclic group and a correspondingly shorter linear chain. The cyclic formed is usually a small group like D3 (hexamethylcyclotrisiloxane), D4 (octamethylcyclotetrasiloxane), or DS (decamethylcyclopentasiloxane), which will then become volatile at coker temperatures. The boiling points of these compounds are listed in Table 1. Several authors have observed that there is not one single breakdown product Breakdown products and boiling points Breakdown species Cyclic D3 Cyclic D4 Cyclic D5 Cyclic D6 Cyclic D7 Cyclic D8 Formula [(CH3)2SiOb [(CH3)2SiO]4 [(CH3)2SiO]5 [(CH3)2SiO]6 [(CH3)2SiO]7 [(CH3)2SiO]s Table 1 66 PTQ SUMMER 2002 Bp @ 760mm ("C) 134 175 210 245 Bp @ 20mm (OC) 154 175 Bp @ 760mm rF) 273 347 410 473 Bp @ 20mm (OF) 309 347 REFINING but rather a range of several small molecules with 03, 04, and OS cyclic oligomers as the most common breakdown products [see references at end]. The new chain end is still subject to further cyclisation and cleaving, so this will continue until the remaining linear species is small enough to be volatile (containing seven or fewer silicon atoms at coker temperatures), or until the remaining material is formed into the coke. This creates an "unzippingll of the polymer chain. Since the reaction will occur on both ends of the polymer chain, for a given weight of material, fewer polymer chain ends means an overall slower rate of degradation. This means that longer initial polymer chains (higher viscosities for rOMS) should lead to greater material life in the coke drum, which will translate into lower use levels. Also, recall that for a defoamer to work, it must be insoluble (but dispersible) in the system. The smaller the chain size, the more soluble rOMS becomes in oil. Therefore, once the polymer falls below some critical minimum chain size (this size will vary with the specific composition of the oil), it will no longer be an effective defoamer. Again, if chain end degradation mechanism holds, longer initial chains are beneficial so that this critical minimum size is reached more slowly. Lab experimental results To support the theoretical degradation hypothesis, TGA and GPC were used to compare the effect of thermal' degradation on two commonly used PDMS viscosities (60000 and 600000cSt) when exposed to a temperature of 400°C (750°F) for two hours at atmospheric pressure under a nitrogen blanket. This is reasonably close to coker conditions except for the lack of oil contact (the coker feed oil would be likely to accelerate the degradation due to some minor catalytic effects, but should not change the mechanism). The PDMS monomer unit is MezSiO, which has a molecular weight of 74. Therefore, the calculated starting average chain lengths are about 1100 for the 600000cSt (60M) and about 2100 for Figure 2 Monthly antifoam usage changing to high viscosity antifoam (600 OOOcSt) the 600M (end group effects on this calculation are minimal). Fora given mass of sample, that means that there are almost twice as many chains, and chain ends, in the lower viscosity material. As previously discussed, chain end cyclisation and cleavageis thought to be the dominant polymer thermal degradation mechanism in this case.We would therefore expect the 60M cSt material to degrade at almost twice the rate of the 600M material, which is exactly what we see in the weight loss numbers (the primary reversion products are volatile at these temperatures). The total weight losseswere 33.7 per cent for the 60M polymer and 18.3 per cent for the 600M polymer during the test (Table 2). This strongly supports the chain end degradation theory. In summary, the weight loss and decreased Mn after heat ageing shows that degradation occurs in both materials, but it occurs more slowly in the higher viscosity material. Also, it is important to note that the molecular weight of the 600 OOOcStis still higher after the thermal exposure than the starting 60000cSt values. This is a good indication that the higher initial molecular weight materials will stay above the critical minimum size for foam control longer, as well as offering lower overall weight loss and bottom line greater effi- Table 2 68 PTQ SUMMER 2002 ciency. The main question, of course, is do these theoretical and laboratory results translate into the field? Field results In working with seveial refineries over the past two decades to control coker antifoam usage, Baker Petrolite has switched to higher viscosity silicone oils in the defoamer formulations. Two examples of this field experience will be given to illustrate the points raised by the thermal stability tests previously described. The first question is: Does use of the higher molecular weight silicone result in less silicone u~agein the coke drum? The graph in Figure 2 shows the actual monthly usage of silicone defoamer on the coker for a North American refinery producing fuel coke. The silicone usage is expressedas the averagenumber of pounds of silicon (Si) added per 1000 barrels of feed to the coker. Each point is a monthly average calculated from actual defoamer inventory reduction during the month and the coker production for the month. The refinery switched from defoamer basedon 60 OOOcSt silicone to one based on 600000cSt between February and March of the first year. The silicone usage dropped dramatically from an average of about 1.4 pounds Si1000bbl to about 0.4 pounds Sij1000/bbl. This is a reduction of 70 per cent in the amount of silicone used to control the foam. Such a reduction has been seen at several other refineries that have made such a switch. The second question is, Does use of the high molecular weight silicone fluids result in reduced silicon contamination of the coker products? To answer that question, look at one North American refinery that recently switched from REFINING 100000cSt based defoamer to a 600 OOOcStbased product. The level of silicon was measured in the coker products using direct injection ion coupled plasma. From previous work at other refineries, Baker Petrolite had determined that the level of silicon in the products could vary by as much as two orders of magnitude, depending on time in the coking cycle that the samples are collected. For this reason, samples were taken just before the drum switch on the last six filling cycles using the 100000cSt based antifoam and the first six filling cycles using the 600 OOOcStbased antifoam. Table 3 is a summary of the trial results. It should be emphasised that the silicon levels determined during the time in the cycle with peak silicone injection are expected to be an order of magnitude greater than the averagevalue over the entire cycle. The highest levels of silicon were seenin the heavy coker naphtha (HCN) but substantial silicon was also found in the light coker gasoil (LCGO) and the heavy coker gasoil (HCGO). This can be explained by the fact that polydimethylsiloxane has multiple breakdown products, primarily the small cyclics resulting from the breakdown as previously described, and each has a different boiling point. The boiling points of several of these breakdown products measured by Patnode are listed in Table 1. Since these boiling points fall throughout the range of typical coker products, any attempt to adjust the endpoints of the coker products in order to produce a silicon-free product cut would not be practical. The second conclusion that jumps out from the data is that the use of higher viscosity silicones in the coke drum can result in dramatic reduction in the silicon contamination of the coker products. Samples taken during other times in the coking cycle were orders of magnitude smaller but also showed similar reductions in the silicon contamination of the coker products. The switch from 100000cSt to 600000cSt silicone fluid was made without any other mechanical or operational changes to the coking process.The time at which the defoamer pumps were started and stopped during the cycle remained the same through the 12 test drums. Thus the silicon reduction can be attributed completely to the change in molecular weight of the coker defoamer. Table 3 degradation studies show that higher viscosity (higher molecular weight) PDMS decomposes more slowly on a weight basis compared to lower viscosity silicone fluids. They are thus expected to retain their foam control characteristics longer than the lower viscosity material, which would translate into a lower dose rate to maintain foam control. Field experience confirms that less high viscosity silicone is required to control the foam in the coke drum and that this results in lower silicon contamination of the coker products and lower overall cost in use, when including the replacement cost of the catalyst. LawrenceN Kremerhasworkedwith BakerPetrolitefor 79 years,and his currrent researchinterestsinclude emulsion technolog)l;defoamersand other aspects of surfacechemistry.He earneda BSin chemistryfrom the Universityof Texasat Austin and a PhD in physical chemistry from the Universityof Utah, followedby a post doctorate at YaleUniversitjt: Timothy G Hueston is senioroil and gas applicationsspecialistfor Dow Corning Corporation,having workedin various functions for the past 18 years.He is currently focusedon the use of silicon-basedmaterialsin oil and gas applications.He holds a BSin chemical engineeringfrom the Massachusetts Institute of Technology. R~ferences Patnode Wand Wilcock D Fi J Am Chem Soc, 68, pp358-363, 1946. GrassieN and MacFarlane I Gi EuropeanPolymerJournal, 14, pp875-884, 1978. Kremer L N, Silicone contamination of coker products; 5th International Conference on Refinery Processing,AIChE Spring National Meeting, 11-14 March 2002. COMPRESS TEMA/ASME Exchanger Vessel Drafter Vessel Coster FEPipe FEA-Nozzles Nozzle Pro These programs design new vessels and heat exchangers, mechanically rerate and evaluate the mechanical integrity of existing vessels and heat exchangers, perform automatic finite element analysis on nozzles with applied loads, perform semiautomatic finite element analysis for a wide variety of vessels and appurtenances, automatically make fabrication and as-built drawings, and perform detailed cost estimates. Avoiding Si contamination Due to the economic loss from silicon poisoning of hydrotreater catalyst, there has been an effort to reduce or eliminate the use of silicon-based defoamers in the coking process. Laboratory thermal Houston, Texas, USA 281-497-5839 FAX www.codeware.com sales@codeware.com 69 PTQ SUMMER 2002