Patent application title: Adaptive Desulfation Control Algorithm
James Edward Mccarthy, Jr. (Canton, MI, US)
Jiyang Yan (Troy, MI, US)
Christian Thomas Chimner (Royal Oaks, MI, US)
Hanlong Yang (Novi, MI, US)
IPC8 Class: AF01N900FI
Class name: Internal combustion engine with treatment or handling of exhaust gas by means producing a chemical reaction of a component of the exhaust gas condition responsive control of heater, cooler, igniter, or fuel supply of reactor
Publication date: 2011-08-04
Patent application number: 20110185708
The exhaust from a diesel-fueled internal combustion engine is treated by
a lean NOX trap. The maximum temperature used for desulfating the
lean NOX trap is kept relatively lower during early life and
progressively increased as the trap ages. Designing for adequate late
life performance entails excess capacity during early life. The method
utilizes the excess capacity available during early life to slow aging of
the trap and thereby extend the trap lifetime. The method facilitates
meeting durability requirements for diesel-powered vehicles with exhaust
1. A method of operating a diesel power generation system in which the
exhaust from a diesel-fueled internal combustion engine is treated using
a lean NOX trap, the method comprising: operating the engine to
produce a lean exhaust comprising NOX and SOX; treating the
exhaust using the lean NOX trap; determining when to denitrate the
lean NOX trap and denitrating the lean NOX trap accordingly;
from time-to-time, desulfating the lean NOX trap by heating the lean
NOX trap to within a desulfating temperature range having an upper
limit temperature and exposing the heated lean NOX trap to rich
conditions under which the lean NOX trap desulfates; aging the lean
NOX trap through many desulfations; and in response to the lean
NOX trap's aging, increasing the upper limit temperature.
2. A diesel power generation adapted, configured, and functional to operate according to the method of claim 1.
3. The method of claim 1, wherein: heating the lean NOX trap to within a desulfating temperature range having an upper limit temperature consists of controlling the heating of the lean NOX trap to bring a characteristic temperature for the lean NOX trap to a predetermined target value; and increasing the upper limit temperature consists of increasing the target value.
4. The method of claim 1, wherein: the aging takes place over two or more consecutive time intervals, each interval comprising at least 1,000 hours of operating the engine and at least 10 desulfations; and the highest upper limit temperature used within each interval after the first is at least 5.degree. C. greater than the highest upper limit temperature used within of the preceding interval.
5. The method of claim 4, wherein the highest upper limit temperature used within the last interval is at least 30.degree. C. greater than the highest upper limit temperatures used within the first interval.
6. The method of claim 4, wherein the highest upper limit temperature within the last interval is at least 50.degree. C. greater than the highest upper limit temperature within the first interval.
7. The method of claim 4, wherein there are three or more of the intervals.
8. The method of claim 4, wherein there are five or more of the intervals.
9. A method of operating a diesel power generation system in which the exhaust from a diesel-fueled internal combustion engine is treated using a lean NOX trap, the method comprising: operating the engine to produce a lean exhaust comprising NOX and SOX; treating the exhaust using the lean NOX trap; determining when to denitrate the lean NOX trap and denitrating the lean NOX trap accordingly; from time-to-time, desulfating the lean NOX trap by heating the lean NOX trap to a desulfating temperature and providing the heated lean NOX trap with a reducing environment; aging the lean NOX trap through many desulfations; and compensating for the effects of the aging by increasing the desulfating temperature.
10. A vehicle comprising a diesel power generation system adapted, configured, and functional to operate according to the method of claim 9.
11. The method of claim 9, wherein the increases to the desulfating temperature are made incrementally and progressively over the lifetime of the lean NOX trap, whereby the lifetime is appreciably extended.
12. The method of claim 9, wherein the desulfating temperature is a main desulfating temperatures and further comprising: between the foregoing desulfations, performing mild desulfations that comprise heating the lean NOX trap to a mild desulfating temperature that is less than the main desulfating temperature and providing the lean NOX trap at the mild desulfating temperature with a reducing environment.
13. The method of claim 12, wherein the mild desulfating temperature is also increased to compensate for the LNT's aging.
14. A method of operating a diesel power generation system in which the diesel exhaust is treated using a lean NOX trap, the method comprising: operating an engine to produce a lean exhaust comprising NOX and SOX; treating the exhaust using the lean NOX trap; from time-to-time, denitrating the lean NOX trap; from time-to-time, desulfating the lean NOX trap by heating the lean NOX trap to within a desulfating temperature range having an upper limit temperature and providing the heated lean NOX trap with reductants; determining the efficacy of the desulfations; and in response to the efficacy determination, increasing the upper limit temperature, whereby the upper limit temperature increases as the LNT ages.
15. A vehicle comprising a diesel power generation system adapted, configured, and functional to operate according to the method of claim 14.
16. The method of claim 14, wherein determining the efficacy of the desulfations comprises evaluating NOX mitigation data collected between desulfations.
17. The method of claim 14, wherein determining the efficacy of a desulfation comprises comparing the performance of the lean NOX trap after the desulfation to the performance of the lean NOX trap before the desulfation.
18. The method of claim 14, wherein determining the efficacy of a desulfation comprises determining whether the desulfation brings the lean NOX trap to a satisfactory level of performance, wherein performance relates to NOX storage capacity or NOX trapping efficiency.
19. The method of claim 14, wherein a duration for the desulfations is adapted according to the efficacy determination and the upper limit temperature is not increased unless the duration is at an upper limit.
20. The method of claim 14, wherein a duration for the desulfations is adapted according to the efficacy determination and the upper limit temperature is not increased unless increasing the duration at the current upper limit temperature has failed to satisfactorily increase the efficacy of the desulfations.
FIELD OF THE INVENTION
 The invention relates to systems having diesel-fueled internal combustion engines with exhaust aftertreatment and methods of operating those systems.
 Diesel-fueled internal combustion engines are used to power vehicles such as medium and heavy duty trucks. Diesel engines are also used in stationary power generation systems. While exhaust aftertreatment systems for gasoline engine have been widely used since the 1970s, diesel engine aftertreatment systems have only recently coming into widespread use.
 Whereas gasoline engines use spark ignition, diesel engines use compression ignition. As a consequence, the composition of diesel exhaust is much different from that of gasoline engines. The major pollutants in gasoline engine exhaust are carbon monoxide, unburned hydrocarbons, and some NOx. The major pollutants in diesel engine exhaust are NOX and particulate matter (soot).
 A catalytic converter, which is an exhaust treatment device comprising a so-called three-way catalyst, can effectively control gasoline engine emissions by oxidizing carbon monoxide and unburned hydrocarbons while also reducing NOX. This approach is unsuitable for diesel engine exhaust because diesel exhaust contains from about 4 to 20% oxygen. The excess oxygen and dearth of oxygen accepting species (reductants) makes catalytic converters ineffective for reducing NOX in diesel exhaust.
 Several solutions have been proposed for controlling NOX emissions from diesel-powered vehicles. One set of approaches focuses on the engine. Techniques such as exhaust gas recirculation and partially homogenizing fuel-air mixtures are helpful, but these techniques alone will not eliminate NOX emissions. Another set of approaches remove NOX from the vehicle exhaust. These include the use of lean-burn NOX catalysts, selective catalytic reduction (SCR) catalysts, and lean NOX traps (LNTs).
 Lean-burn NOX catalysts promote the reduction of NOx under oxygen-rich conditions. Reduction of NOX in an oxidizing atmosphere is difficult. It has proven challenging to find a lean-burn NOx catalyst that has the required activity, durability, and operating temperature range. A reductant such as diesel fuel must be steadily supplied to the exhaust for lean NOX reduction, adding 3% or more to the engine's fuel requirement. Currently, the sustainable NOX conversion efficiencies provided by lean-burn NOX catalysts are unacceptably low.
 SCR generally refers to selective catalytic reduction of NOX by ammonia. The reaction takes place even in an oxidizing environment. The NH3 can be temporarily stored in an adsorbent or ammonia can be fed continuously into the exhaust. SCR can achieve high levels of NOX reduction, but there is a disadvantage in the lack of infrastructure for distributing ammonia or a suitable precursor. Another concern relates to the possible release of ammonia into the environment.
 LNTs are devices that adsorb NOX under lean conditions and reduce and release the adsorbed NOX under rich conditions. An LNT generally includes a NOX adsorbent and a catalyst. The adsorbent is typically an alkali or alkaline earth compound, such as BaCO3 and the catalyst is typically a combination of precious metals including Pt and Rh. In lean exhaust (exhaust containing an excess of oxygen and other oxidizing species in comparison to reducing compounds), the catalyst speeds reactions that lead to NOX adsorption. In a rich exhaust (containing reductants in excess of oxidizing compounds), the catalyst speeds reactions by which reductants are consumed and adsorbed NOX is reduced and desorbed. In a typical operating protocol, a rich condition (reducing environment) is created within the exhaust from time-to-time to regenerate (denitrate) the LNT.
 In addition to accumulating NOX, LNTs accumulate SOX. SOX is the product of combusting sulfur-containing fuels. Even with low sulfur diesel fuels, the amount of SOX produced by combustion is significant. SOX adsorbs more strongly than NOX and necessitates a more stringent, though less frequent, regeneration (desulfation). Desulfation requires elevated temperatures, e.g., 700° C.
 A desulfation process requires much more time than a denitration process. Whereas denitration can be completed in a few seconds, desulfation takes several minutes, commonly on the order of 5-15 minutes. Desulfation could be carried out more rapidly if higher temperatures were used, but normal desulfating temperatures already approach the point at which the LNT will undergo rapid thermal degradation. For example, a temperature of 800° C. may cause a particular LNT to deteriorate and lose a substantial portion of its functionality after a single desulfation.
 U.S. Pat. No. 6,637,198 proposes a desulfation process in which several partial desulfations are performed between each full desulfation. The partial desulfations use lower temperatures and have shorter duration than the main desulfations. The patent asserts that this process facilitates making opportunistic use of higher than normal exhaust temperatures to reduce the amount of fuel expended heating the LNT for desulfations.
 In spite of advances, there continues to be a long felt need for an affordable and reliable diesel exhaust aftertreatment system that is durable, has a manageable operating cost (including fuel requirement), and reduces NOX emissions to a satisfactory extent in the sense of meeting U.S. Environmental Protection Agency (EPA) regulations effective in 2010 and other such regulations that limit NOX emissions from trucks and other diesel-powered vehicles.
 The invention is a method of operating a diesel power generation system in which the exhaust from a diesel-fueled internal combustion engine is treated by a lean NOX trap. The invention extends to encompass systems configured to implement that method. The systems and methods are particularly concerned with how the lean NOX trap is desulfated. According to the invention, the maximum temperature used for desulfating the lean NOX trap is kept relatively lower during early life and increased as the trap ages.
 The storage performance of a lean NOX trap will invariably diminish over time. A lean NOX trap designed to provide adequate late life performance must have excess capacity during early life. The method utilizes the excess capacity available during early life to slow aging of the trap and thereby extend the trap's lifetime. The method facilitates meeting durability requirements for diesel-powered vehicles with exhaust aftertreatment systems.
 In one embodiment, the invention is a method of operating a diesel power generation system in which the diesel exhaust is treated using a lean NOX trap. The engine produces exhaust containing NOX and SOX and the exhaust is treated using the lean NOX trap. From time-to-time, the lean NOX trap is desulfated by heating it to a desulfating temperature, or equivalently, to within a desulfating temperature range. The heated trap is exposed to rich conditions, under which the lean NOX trap desulfates. The lean NOX trap is aged through many desulfations. As the lean NOX trap ages, the highest temperature used for the desulfations is increased. The highest desulfating temperatures used are therefore lower during early lean NOX trap life as compared to mid and late lean NOX trap life.
 The primary purpose of this summary has been to present certain of the inventors' concepts in a simplified form to facilitate understanding of the more detailed description that follows. This summary is not a comprehensive description of every one of the inventors' concepts or every combination of the inventors' concepts that can be considered "invention". Other concepts of the inventors will be conveyed to one of ordinary skill in the art by the following detailed description together with the drawings. The specifics disclosed herein may be generalized, narrowed, and combined in various ways with the ultimate statement of what the inventors claim as their invention being reserved for the claims that follow.
BRIEF DESCRIPTION OF THE DRAWINGS
 FIG. 1 is a flow chart of an exemplary method of adaptive desulfation according to the present invention.
 FIG. 2 is a schematic illustration of an exemplary power generation system that can be configured to operate according to the present invention.
 FIG. 3 is a flow chart of an exemplary method for determining when desulfation is currently required.
 FIG. 4 is a flow chart of an exemplary method for executing a desulfation request.
 FIG. 5 is a flow chart of another exemplary method for adapting desulfating conditions.
 FIG. 1 is a flow chart of an exemplary adaptive desulfation method 100 according to the present invention. The method 100 begins with a determination 101 whether desulfation is required. If desulfation is required the method 100 proceeds to step 102, which analyzes lean NOX trap performance over a preceding interval to determine the effectiveness of prior desulfations using a current set of desulfating conditions. The conditions include a desulfating temperature and optionally additional parameters, duration in this example. If one or more desulfations have been successfully completed using the current desulfating conditions but have failed to produce a satisfactory result, step 103 directs the method 100 to step 105. Otherwise the method 100 proceeds with step 104 and carries out a desulfation using the extant conditions.
 Step 105 determines whether measures can be taken to improve desulfation without increasing the desulfating temperature. In this example, step 105 determine whether the desulfating time currently in use is less than a maximum. If not, the method 100 increases the desulfating time and proceeds with the desulfation. If the desulfating time is already at a maximum, the method proceeds to step 107.
 Step 107 determines whether the desulfating temperature is already at a limit. The method 100 raises the desulfating temperature over the lifetime of a lean NOX trap, but still uses a lifetime limit on how high the desulfating temperature can go. If the limit has been reached, a fault is indicated in step 110 and the desulfation is carried out without further increasing the severity of the desulfating conditions. If the limit has not yet been reached, the method 100 proceeds with step 108 in which the desulfating temperature is increased. The desulfating time is reduced in step 109 and the desulfation carried out in step 104.
 The method 100 will increases the desulfating time again, later, as necessary, while retaining the increased desulfating temperature. Further increases to the desulfating temperature will be deferred until at least one successfully completed desulfation using the higher desulfating and the maximum desulfating time has proven inadequate. Optionally, decreases to desulfation time can be made between steps 103 and 104, whereby the duration of desulfation is adapted either upwards or downwards according to desulfation performance as determined from the data analyzed in step 102. Increases to the maximum temperature used for desulfation, however, are preferably monotonic.
 FIG. 2 provides a schematic illustration of an exemplary power generation system 200 that can be configured, adapted, and functional to implement methods of the invention. The exemplary power generation system 200 can be installed in a vehicle or in a stationary power application, such as a back-up generator for a hospital. While the invention may be described and claimed in terms of power generation systems, it should be appreciated that the full value of the invention may be realized in a larger system, such as vehicle. For example, the invention extends the period over which a vehicle can operate within the confines of emission control regulations without requiring replacement of any aftertreatment system components. This can be critical in meeting regulatory and customer requirements.
 The exemplary power generation system 200 comprises an engine 201, a manifold 202, and an exhaust aftertreatment system 203. The exhaust aftertreatment system 203 comprises an exhaust line 204 configured to channel exhaust from the manifold 202 through, in order, a fuel reformer 205, a thermal mass 206, an LNT 207, a DPF 208, and an SCR catalyst 209. A fuel injector 211 is configured to inject fuel into the exhaust line 204 upstream from the fuel reformer 205 at times and at rates determined by the controller 210. Implementation of the present invention does not require either the fuel reformer 205, the thermal mass 206, the DPF 208, the SCR catalyst 209, and the fuel injector 211. The major device that relate to the invention are the engine 201, the LNT 207, and the controller 210.
 The controller 210 may be a control unit for the engine 201 or a separate control unit. If separate, the controller 210 preferably communicates with the engine control unit. Configuring the system 200 to practice a method of the invention generally involves providing the controller 210 with suitable programming. With suitable programming and any other necessary adaptations, the system 200 will be functional to carry out the method.
 The controller 210 receives data from various sensors, such as a temperature sensor 212. The sensor 212 is configured to sense a characteristic temperature for the LNT 207. Other sensors that may be provided include, without limitation, a temperature sensor for the fuel reformer 205 and one or more exhaust composition sensors that can be used to monitor performance of the LNT 207. Usually there will be at least one composition sensor downstream from the LNT 207, such as a NOX sensor. Suitable locations include locations upstream and downstream from the SCR catalyst 209.
 The engine 201 can be any engine that operates to produce a lean exhaust stream comprising NOX and SOX. Generally the engine 201 is a diesel-fueled compression ignition internal combustion engine that produces an exhaust containing from 2 to 20% oxygen. The diesel exhaust is typically at temperatures in the range from about 200 to about 500° C., with temperatures in the range from 250 to 450° C. beginning common. The manifold 202 couples the exhaust aftertreatment system 203 to an exhaust stream from the engine 201. Preferably the exhaust system 203 comprises a single exhaust line 204 that receives the entire exhaust from the engine 201.
 The exhaust aftertreatment system 203 and the exhaust line 204 preferably have no valves or dampers that control the flow of exhaust. Exhaust system valves and dampers provide control over the distribution of exhaust between a plurality of flow paths. Such control is desirable in terms of limiting fuel usage. Reducing the flow of exhaust to the fuel reformer 205 and the LNT 207 during rich regeneration would reduce the amount of fuel expended eliminating oxygen from the exhaust in order to provide rich conditions. The reduced flow rate would also increase residence times, and thus the efficiency with which reductants are used. Nevertheless, it is preferred that the exhaust treatment system 203 operate without exhaust line valves or dampers in order to avoid failures resulting from reliance on such devices.
 The LNT 207 is a device that adsorbs NOX under lean conditions and reduces NOX releasing the reduction products (N2 and NH3) under rich conditions. Some alternate terms for a lean NOX trap (LNT) are NOX absorber-catalyst and NOX trap-catalyst. An LNT generally comprises a NOX absorbent and a precious metal catalyst in intimate contact on an inert support. Examples of NOX adsorbent materials include certain oxides, carbonates, and hydroxides of alkaline earth metals such as Mg, Ca, Sr, and Ba or alkali metals such as K or Cs. The adsorption can be physical or chemical, but is generally primarily chemical. The precious metal typically comprises one or more of Pt, Pd, and Rh. The support is typically a monolith, although other support structures can be used. The monolith support is typically ceramic, although other materials such as metal and SiC are also suitable for LNT supports. The LNT 207 may be provided as two or more separate bricks.
 The fuel reformer 205 and the fuel injector 211 are part of a system for producing the rich conditions and providing the reductant required for denitration and desulfation. A reductant is a compound that is reactive to accept oxygen and become oxidized. The reductant is generally diesel fuel or a substance derived from diesel fuel by partial combustion and or steam reforming reactions. A rich condition for the exhaust is one in which the concentration of reductants is more than stoichiometric for combustion with any oxygen and other oxidizing compounds present. In other words, a rich environment is one in which there is an excess of reductant and the overall composition is reducing rather than oxidizing.
 Optionally, the engine 201 is used to assist in producing rich conditions. If the engine 201 can be operated with rich combustion or with post-combustion fuel injection, than the engine 201 can provide a rich mixture and the exhaust line fuel injector 211 is optional. The engine 201 can also facilitate generating rich conditions by measures that reduce the exhaust oxygen flow rate. Such measures may include, for example, throttling an air intake for the engine 201, increasing exhaust gas recirculation (EGR), modifying cylinder injection controls, and shifting gears to reduce the engine speed.
 It is preferable for the aftertreatment 203 to be capable of providing the rich conditions for denitration and desulfation of the LNT 207 while making few or no changes to the operation of the engine 201 in order to avoid having regenerations (denitrations and desulfations) adversely affect drivability and also to provide greater independence between the designs and configurations of the aftertreatment system 203 and the engine 201. It would not be unusual for the engine 201 to be manufactured by one company while the power generation system 200 comprising the engine 201 is assembled by another company. A third company may build a vehicle using the assembled power generation system 200.
 The fuel reformer 205 is a device that is functional to reform diesel fuel into reformate, especially CO and H2. Reformate is a better reductant than diesel fuel for denitrating the LNT 207. Reformate is more reactive than diesel fuel and results in less NOX slip. NOX slip is the release of unreduced NOX from the LNT 207 during denitration.
 Preferably, the fuel reformer 205 has a low thermal mass and comprises both oxidation and steam reforming catalysts. A low thermal mass allows the fuel reformer 205 to be heated to steam reforming temperatures for each denitration without requiring an excessive amount of time or fuel. Steam reforming temperatures have a minimum in the range from about 500 to about 600° C., typically requiring at least 550° C. At steam reforming temperatures, energy from oxidation and partial oxidation, which are exothermic, can drive steam reforming, which is endothermic. This improves the efficiency with which reformate is produced and decreasing the amount of waste heat. A sufficiently low thermal mass can be achieved by constructing the fuel reformer 205 around a monolith substrate formed of thin metal foils, e.g., 130 microns or less. Preferably the foils are 100 microns or less, and more preferably 50 microns or less. The preferred structure can be heated from a typical diesel exhaust temperature in the range from 250 to 300° C. to steam reforming temperatures in 2 or 3 seconds or less.
 The exhaust from the engine 201 generally comprises at least 2% oxygen. When fuel is added to the exhaust to produce a rich condition for denitration or desulfation, this oxygen is eliminated by combustion. In the system 200, this combustion takes place in the fuel reformer 205. If the combustion does not take place upstream from the LNT 207, it will generally take place within the LNT 207. The precious metal catalysts typically used by the LNT 207 are functional as oxidation catalysts. If too much combustion takes place within the LNT 207, it can cause undesirable temperature excursions which are particularly problematic if they take place during denitrations. Such temperature excursions can cause wear and result in the release of unreduced NOX.
 Optionally, a burner or any device that is functional to bring about combustion, such as an oxidation catalyst, a three-way catalyst, or a suitably catalyzed diesel particulate filter can be used instead of the fuel reformer 205. Like the fuel reformer 205, these device can cause combustion to take place upstream from the LNT 207. They may also accomplish a certain amount of fuel reformate through partial oxidation reactions.
 When combustion takes place upstream from the LNT 207 in preparation for or during denitration, the heat is preferably held temporarily within devices upstream from the LNT 207 to be released only slowly over a prolonged period. According to the preferred design, the fuel reformer 205 has a low thermal mass (thermal inertia) and is not very effective for holding heat. In the system 200, the thermal mass 206 provides the desired heat retention function.
 The thermal mass 206 is any device that is effective for exchanging heat with the exhaust and storing the heat produced by the fuel reformer 205 over the course of a denitration without heating excessively. A suitable device can be simply a catalyst substrate, with or without a catalyst. A suitable device is, for example, an inert monolith substrate, either metal or ceramic. Preferably, the thermal mass 206 has a thermal inertia that is greater than that of the fuel reformer 205. The DPF 208 can be used as the thermal mass 206, although in the exemplary system 200, the DPF is downstream from the LNT 208 and instead serves to help protect the SCR catalyst 209 from high temperatures during desulfations.
 The DPF 208 and the SCR catalyst 209 contribute to meeting emission control limits and durability requirements. The DPF 208 removes particulate matter from the exhaust, which is the major pollutant in diesel exhaust other than NOX. The SCR catalyst 209 provides supplementary NOX mitigation. It improves durability by allowing sufficient NOX mitigation to be maintained with less frequent denitration and desulfation of the LNT 208. When some NOX is reduced downstream from the LNT 207, the LNT 207 does not need to be maintained at as high a level of efficiency.
 A DPF is a device that traps particulates matter (soot), removing it from the exhaust flow. The DPF 208 can be a wall flow filter, which uses primarily cake filtration, or a flow-through filter, which uses primarily deep-bed filtration. The DPF 208 can have any suitable structure. Examples of suitable structures include monoliths. A monolith wall flow filter is typically made from a ceramic such as cordierite or SiC, with alternating passages blocked at each end to force the flow through the walls. A flow-through filter can be made from metal foil.
 Trapped soot can be removed from the DPF 208 continuously by catalyzing reactions between soot and NOX, but typically the DPF 208 must be heated from time-to-time to a temperature at which it regenerates by combustion of trapped soot. The temperature required for soot combustion can be reduced by a catalyst. Suitable catalysts include precious metals and oxides of Ce, Zr, La, Y, and Nd. Soot combustion is exothermic and can be self-sustaining once ignited.
 The SCR catalyst 209 is an ammonia-SCR catalyst functional to catalyze reactions between NOx and NH3 to reduce NOx to N2 in lean exhaust. Examples of SCR catalysts include oxides of metals such as Cu, Zn, V, Cr, Al, Ti, Mn, Co, Fe, Ni, Pd, Pt, Rh, Rd, Mo, W, and Ce, zeolites, such as ZSM-5 or ZSM-11, exchanged with metal ions such as cations of Cu, Co, Ag, Zn, or Pt.
 The engine 201 operates to produce a lean exhaust comprising NOX, particulate matter, and SOX. The expression NOR designates the family of molecules consisting of nitrogen and oxygen atoms, primarily NO and NO2. The subscript X indicates the family includes multiple species with varying proportions between nitrogen and oxygen atoms. The notation SOX is similar.
 Under lean conditions, the LNT 207 absorbs a portion of the NOX and a portion of the SOX in the exhaust. If the SCR catalyst 209 contains stored ammonia, an additional portion of the NOX is reduced therein. The DPF 208 removes at least a portion of the particular matter from the exhaust.
 From time-to-time, the controller 210 determines to denitrate the LNT 207. For denitration, the fuel reformer 205 is heated to steam reforming temperatures by injecting fuel into the exhaust line 204 through the fuel injector 211 under the control of the controller 210 at rates that leave the exhaust lean. Under lean condition, most of the injected fuel combusts in the fuel reformer 205, heating it. After the fuel reformer 205 has reached steam reforming temperatures, as may be determined using a temperature sensor, the fuel injection rate is controlled to make the exhaust condition rich for a period of time (rich phase) over which the LNT 207 denitrates.
 During the rich phase, injected fuel mixed with lean exhaust enters the fuel reformer 205. A portion of the fuel combusts, consuming most of the oxygen from the exhaust. Another portion of the fuel is converted to reformate (syn gas), which is primarily H2 and CO. The reformate enters the LNT 207 where it reacts to reduce and release trapped NOX. Most of the NOX released during the rich phase is reduced to N2 or NH3, although it is typical for a small amount to be released (slip) without being reduced. The NH3 is mostly trapped by the SCR 209, where it is generally consumed reducing NOX over the course of the following lean phase.
 NOX slip occurs primarily at the beginning of the rich phase and may be lessened by varying the reductant concentration over the course of the denitration. The preferred reductant concentration profile has the reductant concentration relatively low at the start of the rich phase and gradually increasing over at least a first portion of the rich phase.
 When the fuel reformer 205 heats for denitration, the thermal mass 206 also heats, but to a lesser degree. After denitration is complete, fuel injection ceases and the fuel reformer 205 cools down to exhaust temperatures. The thermal mass 206 also cools. The LNT 207 will heat over the course of the denitration and a short period following, but only to a modest degree.
 From time-to-time, the LNT 207 must be heated substantially in order to carry out a desulfation. In the system 200, the LNT 207 can be heated by injecting fuel into the exhaust. The injected fuel combusts, primarily in the fuel reformer 205. Over the course of a few minutes, the thermal inertias of the thermal mass 206 and of the LNT 207 are overcome and the LNT 207 reaches desulfating temperatures. Alternative and supplemental means of heating the LNT 207 include, without limitation, engine measures, such as operating the engine to produce a hot exhaust, burners, and electrical heaters.
 In the system 200, after the LNT 207 has reached desulfating temperatures, the fuel injection rate is controlled to make the exhaust rich. It might be considered ideal to maintain the rich condition until the LNT 207 has desulfated to a desired degree. In the system 200, however, it proves difficult to continuously maintain rich conditions while also maintaining the fuel reformer 205 and the LNT 207 within desired temperature ranges. In general, it is necessary to pulse the fuel injection over the course of a desulfation.
 In the context of maintaining the desired conditions for desulfation, pulsing the fuel injection means creating alternating rich and lean phases (periods). During the rich phases, the reformer 205 heats and produces reformate. During the lean phase, fuel injection ceases and the fuel reformer 205 cools. Typically, the durations of the rich phases are in the range from about 4 to about 30 seconds, with periods in the range from about 5 to about 15 seconds being preferred.
 When lean and rich phases are alternated in this manner, some combustion takes place in the LNT 207. The LNT 207 typically comprises oxygen storage materials, which are materials that are functional to store oxygen. These materials accumulate oxygen during the lean phases, typically to the point of saturation. During the rich phases, the reductants react with the stored oxygen, consuming the stored oxygen and producing heat. The amount of heat can be significant, e.g., enough to make the LNT 207 50 to 100° C. hotter than the fuel reformer 205. The amount of this heating is generally proportional to the oxygen storage capacity of the LNT 207 and to the frequency of switches between lean and rich phases.
 In a preferred embodiment, the fuel injection rate is selected to provide a desired reformate production rate or concentration while the temperatures of the fuel reformer 205 and the LNT 207 are controlled by varying two parameters that dictate the lean-rich pulse pattern. The lean rich pulse pattern consists of the lean phase lengths and the rich phase lengths or two other parameters that define these, such as the pulse frequency and the ratio between lean time and rich time.
 In one exemplary control strategy, the rich phases are terminated when the fuel reformer 205 reaches a pre-defined upper limit temperature, such as 650° C. The lean phase durations are then adjusted in a closed loop control algorithm to maintain the temperature of the LNT 207 within the desired temperature range. In a variation of this method, the adjusted parameter is a temperature to which the fuel reformer 205 is required to cool before terminating. The LNT temperature target by control is preferably a maximum temperature that the LNT reaches over the course of a lean-rich cycle, but could alternatively be another temperature, such as an average temperature. Making the pulse periods shorter by reducing the durations of the lean phases raises temperatures within the LNT 207. The temperatures rise because the fuel reformer 205 is on average hotter resulting in more heat convection to the LNT 207 and because the more frequent alternation between lean and rich phases results in more combustion within the LNT 207. Conversely, lengthening the lean phases lowers temperatures within the LNT 207.
 In another exemplary control strategy, the durations of the rich phases are predetermined and both the upper and lower temperatures of the fuel reformer 205 (or two equivalent parameters) are set in order to achieve the desired rich phase length while maintaining the LNT 207 at the desired temperature. In this example, the temperature to which the fuel reformer 205 falls during a lean phase is raised or lowered to raise or lower the temperature of the LNT 207. During the rich phases, the reformer heats by an amount that depends on the predetermined rich phase duration. By selecting the rich phase duration within suitable limits, the fuel reformer 205 can be prevented from either overheating or cooling excessively, e.g., cooling below steam reforming temperatures. Preselecting the rich phase duration can be beneficial in managing hydrocarbon emissions during desulfations. In terms of limiting hydrocarbon emissions, a suitable length for the rich phases is on the order 10 to 20 seconds for a fresh catalyst, decreasing to about 50-70% as much as the catalyst ages.
 It is generally also necessary to regenerate the DPF 208 from time-to-time. Regenerating the DPF 208 comprises heating the DPF 208 to temperatures at which soot trapped with the DPF 208 combusts. The DPF 208 can be heated in the same way as the LNT 207 is heated for desulfation. Soot combustion is generally self-sustaining. Once the DPF 208 is heated to soot combustion temperatures, it is generally not necessary to supply any additional heat. The upstream devices, including the fuel reformer 205 and the LNT 207, can be allowed to cool while soot combustion is proceeding to completion.
 The DPF 208 is typically of the wall flow filter variety and must be regenerated often enough to avoid excessive back pressure. Ideally, regenerating the DPF 208 each time the LNT 207 is regenerated provides sufficient frequency. The DPF 208 is heated to soot combustion temperatures each time the LNT 207 is heated for desulfation. If the DPF 208 has sufficient capacity to require regeneration no more often than the LNT 207 is regenerated, supplemental fuel expenditure and additional heating of the LNT 207 for the sole purpose of regenerating the DPF 208 can be avoided.
 One approach that can facilitate not having to regenerate the DPF 208 more often than the LNT 207 is desulfated is to provide a second DPF downstream from the fuel reformer 205 and upstream from the LNT 207. This second DPF can be used as the thermal mass 206, but is preferably a low thermal mass device upstream from the thermal mass 208. Preferably, this second DPF is of the flow-through type. Preferably, its thermal mass is sufficiently low that it heats and regenerates each time the fuel reformer 205 is heated to supply reformate for denitration. Accordingly, in this embodiment, there is a second DPF that regenerates as often as the LNT 207 is denitrated.
 The times at which the LNT 207 is denitrated are determined by the controller 210 and can be determined in any suitable manner. Typically, certain threshold must be met before allowing a denitration to begin. Threshold criteria can be, for example, one or both the LNT 207 and the fuel reformer 205 being at minimum temperatures, the oxygen concentration being below a maximum (e.g., less than 15%), the flow rate being above a minimum (e.g., significantly greater than at idle), the engine speed variance, as determined by a moving average, being below a maximum, at a minimum time elapsed since the last denitration, and a gear shift not currently imminent or in progress. If the threshold criteria are met, regeneration will begin if an additional condition (or conditions) are met. An additional condition generally relates to a measure of how urgently denitration is needed and is optionally weighed against the suitability of current conditions for beginning a denitration.
 The urgency of the need to denitrate generally relates to one or more of NOX loading of the LNT 207, remaining NOX storage capacity, NOX trapping efficiency (optionally normalized for such factors as the LNT temperature and exhaust flow rate), NOX concentration in the exhaust at a point downstream from the LNT 207, and cumulative NOX emissions since the last denitration (optionally normalized by the engine's toque production). A measure of suitability can relate to one or more of such factors as the exhaust oxygen concentration (low is preferred), the engine speed variance (low is preferred), and the exhaust flow rate. One procedure for weighing the urgency of the need to denitrate against the suitability of current conditions to denitration is to assign numerical values to the urgency and the conduciveness, multiplying the two together, and denitrating based on whether the result exceeds a predetermined critical value.
 Likewise, the times at which the LNT 207 is desulfated are determined by the controller 210 in any suitable manner. Threshold criteria may be employed similar to those used for denitration and a measure of suitability of current conditions to desulfation can be weighed against a measure of the urgency of the need for desulfation. The urgency of the need to desulfated can be based on, for example, one or more of an estimate of the amount sulfur trapped in the LNT, the frequency with which denitration is being required, an estimate of the post-denitration NOX storage capacity of the LNT 207, the amount of time since the last desulfation, the number of denitrations since the last desulfation, and an estimate of the average LNT efficiency following the last desulfation, or over the last several desulfations. The LNT efficiency can be normalized to separate changes intrinsic to the LNT 207 from changes in the operating regime of the engine 201. Alternatively, normalization can be limited, whereby less sulfur loading is tolerated when the engine is in an operating regime that requires peak LNT efficiency, e.g., when the engine 201 is in a high speed-high load condition. Preferably, the determinations of when to desulfate the LNT 207 include a dynamic measure of LNT performance, whereby adjustments to the desulfation timing and the desulfating conditions can be adapted to measurable indications of aging.
 FIG. 3 provides a flow chart of an exemplary method 300 that determines whether there is a need to desulfate the LNT 207. After initialization 301, the method 300 begins with a threshold determination 302. In this example, the threshold determination is whether a minimum number of denitrations, A, have taken place since the last desulfation. The use of a threshold determination provides a failsafe for preventing overly frequent desulfations. Alternate or additional criteria can be used to make this threshold determination, such as whether a minimum time of operation has elapsed since the last desulfation.
 If the threshold criteria are satisfied, the method 300 sets a desulfation request flag in step 306 if any of several criteria are met. These criteria are tested through a series of steps, 303-305. The first criteria examines the median of a normalized LNT efficiency over the last several lean-rich cycles of NOX trapping followed by denitration. The efficiency is normalized for the LNT temperature and the exhaust flow rate to provide a value that is largely independent of the engine 201's speed and load. If the efficiency is below a threshold value B, the desulfation request flag is set. Step 304 checks the elapsed operating time (engine running hours) since the last desulfation. An alternative criteria could be based on the number of denitrations since the last desulfation. Step 305 checks whether the SOX loading is greater than a critical value D.
 The SOX loading is estimated. The estimate generally takes into account at least the amount of fuel used since the last desulfation and the estimated sulfur content of that fuel. The accumulation rate is optionally modified by an accumulation efficiency that depends on the temperature of the LNT 207 and or the exhaust flow rate. Optionally, the estimate includes an amount remaining after the last desulfation, which would be particularly relevant if the last desulfation was aborted prematurely.
 If the method 300 determines there is a need for desulfation, the desulfation request is set in step 306. The actual start of the desulfation may be postponed until conditions are suitable, but a desulfation will begin in response to the determination. A separate algorithm checks the suitability of current conditions, The method 300 is executed periodically, e.g., after each denitration, within the course of a more broadly functioning control algorithm. If the desulfation request flag is set in step 306 before the routine exists through the return step 308, the parameters for desulfation are set in step 307.
 The parameters for desulfation include a desulfating temperature. Another parameter is typically set to determine a duration for the desulfation. This could be, for example, a total time at rich conditions, a total amount of reductant to be provided to the LNT 207 at desulfating temperature, or a total amount of sulfur to be removed. The later envisages a dynamic determination of the sulfur removal rate as a function of measured values such as the LNT temperature. Another parameter that may be set is a duration for individual rich phases to be used over the course desulfation.
 The example shows the determination of parameters for desulfation being made immediately after setting the desulfation request flag. Optionally the selection of parameters is postponed until it has been determined that conditions are suitable for beginning the desulfation. Postponement in this manner allows a desulfation duration parameter to be adjusted to account for additional sulfur accumulation that may occur between setting the desulfation request flag and finding that conditions are suitable to begin desulfating.
 Step 307, selecting the parameters to use for the desulfation, invokes another method. This method is exemplified by the process 500, which is illustrated with a flow chart in FIG. 5. The method 500 selectively modifies the desulfation parameters from previously used values based on how efficacious the previous desulfations have been in improving the performance of the LNT 207.
 After initialization 501, the first step in the method 500 is to determine 502 whether the last desulfation was completed successfully. This determination of whether a desulfation was successful may consider factors in addition to whether the desulfation was completed. Examples of additional factors are the number of rich pulses required to complete the desulfation, the median rich pulse duration, and the mean LNT temperature during the rich pulses. The desulfation may be considered unsuccessful if the number of rich pulses required to make up the rich time was excessive, if the median rich pulse duration was too short, or if the mean LNT temperature was too far below the intended temperature range. Only desulfation qualified as successful are considered in deciding whether to adapt the desulfation conditions. Step 502 prevents any adjustment to the current desulfating parameters before they have proven inadequate to restore the LNT 207 to effectiveness. If one or more desulfations using the current desulfating conditions have been completed successfully, the method 500 proceeds with step 503, in which the efficacy of the last desulfation, or the last several desulfation, in improving LNT performance is evaluated.
 The assessment of LNT performance improvement can be made in any suitable manner. Any suitable measure of LNT performance can be used. In one example, LNT performance is determined according to a normalized LNT efficiency averaged over a period following the last desulfation. The efficiency is preferably measured with input from a NOX sensor within the exhaust line 204 at a position downstream from the LNT 207. The average is preferably taken over at least several cycles of NOX trapping followed by denitration. As another example, the LNT performance can be determined based on how frequently it has been necessary to denitrate the LNT in order to meet pollution control targets. If a target degree of pollution control can be maintained with less frequent denitration, the LNT is performing better.
 The determination of efficacy can be made by evaluating the LNT performance in either relative or absolute terms. An example of a relative determination is one comparing the average LNT efficiency over a period before the last desulfation to an average LNT efficiency over a period following the last desulfation. In that case, the efficacy determination is based on the degree of improvement in performance. An example of an absolute determination is one comparing the LNT efficiency following the last desulfation against a fixed reference. A desulfation is effective if it improves LNT performance to a satisfactory degree.
 If the previous desulfation was not sufficiently effective, step 504 determines whether a desulfation duration parameter is already raised to its upper limit. When desulfation with current parameters is not producing a satisfactory result (is not proving sufficiently effective), the method 500 responds by increasing desulfating time before increasing the desulfating temperature. The method 500 uses a predetermined maximum, which can be a fixed value or a function of a parameter that correlates to the LNT's aging, such as the desulfation temperature. If the maximum desulfation time is variable, then it preferably diminishes as the LNT ages. An aged LNT has less functional storage capacity that is amendable to restoration by desulfation. Also, desulfation proceeds more quickly at higher temperatures.
 As an alternative to using a predetermined maximum duration, step 504 can instead analyze LNT performance data to determine whether the last increase in desulfation time produced a satisfactory degree of improvement in desulfation efficacy. In this alternative, the desulfation time is increased as required until the data shows these increases have reached a point of diminishing returns. In any event, if further increases to the desulfation time are tenable, because an upper limit or point of diminishing returns has not yet been met, then the step 504 directs an increase in desulfation time, step 511. Otherwise the method proceeds with step 505.
 Step 505 avoids premature increases to the desulfating temperature by preventing any increase unless desulfation using the current desulfating temperature and the maximum desulfating time has proven unsatisfactory through several attempts. Increases to the maximum temperature used for desulfation are preferably made monotonically. Also, the increases are preferably made only when it is clear that the LNT 207 cannot be operated satisfactorily without employing at least some desulfations at an increased temperature. These preferences provide the greatest deferment of LNT aging. If step 505 has been reached several times in a row due to several desulfations using the current desulfating temperature and maximum time failing to provide a satisfactory result (as determined by steps 503 and 504), then the method 500 proceeds with step 506, from which the desulfating temperature can be raised. Otherwise, the method 500 proceeds with step 513, which directs that the parameters for the last desulfation be used again. Steps 503 and 513 are no different. They are illustrated separately to make the process flow easier to display.
 Step 506 makes a final check before raising the desulfating temperature. This step ensures that the desulfating temperature is never raised above a predetermined upper limit. The upper limit temperature is set at the point where the deterioration experienced by the LNT is likely to outweigh the benefits of deeper desulfation regardless of how much the LNT 207 has already aged.
 A typical upper limit is the one specified by the LNT manufacturer, or a slightly higher temperature. While an aged LNT may be less vulnerable than a fresh LNT to the effects of high desulfating temperatures, the primary mechanism of the invention is to take advantage of overdesign. A fresh LNT is overdesigned in anticipation of diminishing performance over time. Overdesign for early life is necessary to ensure satisfactory late life performance. The invention takes advantage of this overdesign to use lower desulfating temperatures during early life and thereby slow the process of aging. The diminished effectiveness of low temperature desulfations is tolerated as long as possible in order to maintain functionality over a longer period.
 If the upper limit temperature has not yet been reached, then the method 500 proceeds with step 511 in which the desulfating temperature is raised. The magnitudes of the increases made in step 511 can be chosen in any suitable manner. Generally the magnitudes of the increases in desulfating temperature are predetermined. Examples of predetermined increases include, for example 5° C. or 10° C. each time the desulfating temperature is raised. Any suitable series of preselected temperatures can be used. Another example provides a series of temperatures that rise linearly on a logarithmic scale from the lowest desulfation temperature, used for a fresh catalyst, to the upper limit temperature. The upper limit temperature is not used until the LNT 207 has aged into mid or late life.
 If step 506 is reached with the upper limit desulfating temperature already in use, then the method proceeds with step 512, Step 512 signals a fault and the need for service. When this fault condition is reached, operation may continue as shown. While the desulfations may be unsatisfactory, the aftertreatment system 203 may remain functional albeit outside specification in terms of either unsatisfactory pollution control or excessive fuel usage. The system 200 may begin denitrating very frequently in attempts to maintain the overly aged LNT at a satisfactory level of NOX trapping efficiency. To limit this type of behavior, step 512 may implement other fault procedures. One option is to terminate the LNT regeneration cycle entirely until the unit has been serviced.
 FIG. 4 provides a flow chart of an exemplary method 400 in which a desulfation is carried out using the selected desulfating parameters. The method 400 is executed periodically whenever the desulfation request flag is set. For example, it might be executed after each denitration, or every few seconds. Following initialization, the first step in the method 400 is to determine whether conditions are suitable for starting a desulfation. Similar to the threshold determination for beginning a denitration, step 402 can require one or more of a minimum temperature for the LNT 207 or the fuel reformer 205, a maximum exhaust oxygen concentration, a minimum exhaust flow rate, a maximum engine speed variance as determined by a moving average, and a check that a gear shift is not currently in progress. Optionally, the conduciveness of conditions to beginning a desulfation is weighed against the time since the desulfation request flag was set, or a similar criteria relating to the urgency of the need to desulfate, whereby more flexible criteria such as the engine speed variance can be given greater latitude over time in order to avoid unnecessary and excessive waiting for preferred conditions to arrive.
 Once step 402 determines that conditions are satisfactory, the method 400 proceeds with step 403 which checks the soot loading level of the DPF 208. The amount of soot can be determined in any suitable manner. One option is to calculate the amount, based for example on the engine's speed load history since the last DPF regeneration (DeSoot operation) in combination with knowledge of the engine 201's particulate matter production rate as a function of speed and load. Another option is to determine the amount of soot from the pressure differential across the DPF, the temperature of the DPF, and the engine exhaust flow rate. If it is determined that too much soot has accumulated, then step 404, a special DeSoot operation 404 is performed before proceeding to heat the LNT 207 to desulfating temperatures in step 405.
 The special DeSoot operation 404 is performed to avoid overheating the DPF 208. When the DPF 208 is heavily loaded with soot, there is a danger that once heating initiates soot combustion, the combustion will further heat the DPF 208 to a point where damage takes place. To minimize this risk, step 404 ceases the provision of supplemental heating once the DPF 208 is hot enough for soot combustion to be underway. After the DPF 208 has had time to regenerate, the step 404 allows the method 404 to resume heating for desulfation with step 405.
 Step 405 checks whether the LNT 207 is at a suitable temperature to begin the preferred cycles of alternating lean and rich phases to provide temperature control and desulfation conditions as described above. Step 405 may also check the temperature of the fuel reformer 205. If the LNT 207 is not yet hot enough, the method 400 proceeds with step 406, which heats the LNT 207. Heating may be accomplished in any suitable manner. For example, fuel may be injected into the exhaust line 204 at a rate where the fuel combusts within the reformer 205 to raise it to its maximum operating temperature and then maintain the reformer 205 at that temperature until the LNT 207 is adequately heated.
 Once the LNT 207 is adequately heated, the method 400 proceeds with step 407 in which desulfation proceeds. In this example, desulfation proceeds through pulsed fuel injection as described above. In this context, pulsed fuel injection alternates periods of fuel injection that creates rich condition with period of no fuel injection that leave the exhaust lean. Pulsing in this context should not be confused with pulse width modulated flow control, which may be used by the fuel injector 211 to provide a desired fuel injection rate. Pulse width modulated flow control has a high frequency and results in an essentially constant fuel dosing rate as opposed to low frequency pulsing which creates alternating lean and rich conditions within the exhaust line 204.
 After each rich phase or with other suitable timing, the method 400 proceeds to step 408 in which a progress variable for the desulfation is advanced. The progress variable can be of the types described previously. For example, the progress can be measured by the accumulated time at which the LNT 207 has been at rich conditions or the amount of reductant that has been provided to the LNT 207 at desulfating temperature. After advancing the progress variable, the method 400 proceeds with step 409 in which the value of the progress variable is tested to determine whether desulfation is complete. In this example, the total time at rich conditions is compared to the desulfation time set by the method 500.
 If desulfation is complete, the desulfation request flag is reset (turned off) in step 411 and the process 400 completes. If, not the method 400 proceeds to step 410, which determines whether conditions remain suitable for desulfation. This determination can be similar to step 402, but generally with less stringent criteria. The criteria ensure that the exhaust conditions are amendable to controlling temperatures in the reformer 205 and the LNT 207.
 The method 400 illustrates only one of several possible procedures that may be followed if conditions become unsuitable for continuing a desulfation that is in progress. The procedure shown has the desulfation aborting and resetting the desulfation request flag. Optionally, the desulfation flag can remain set, whereby desulfation resumes once conditions are again suitable. Optionally, step 410 waits for a period to determine if the problematic condition is fleeting before aborting. During this waiting period, fuel reformer and LNT temperatures can be maintained by fuel injection. The procedure taken can be made dependent on the condition that caused the interruption. For example, if the problem condition is excessive engine speed variance, it is likely to pass soon and waiting is preferred. If the problem condition is idle (assuming that desulfation at idle is problematic in the system 200), then aborting the desulfation may be the better option.
 Each complete desulfation takes from about 3 to about 30 minutes, more typically 5 to 10 minutes, e.g., 7.5 minutes. Shorter desulfations are generally avoided because it usually takes several minutes to heat the LNT 207 to desulfating temperatures. Longer desulfation is generally unnecessary.
 Desulfation comprises heating the LNT 207 to a desulfating temperature, or equivalently, to within a range of desulfating temperatures. A desulfating temperature refers to any characteristic temperature for the LNT 207. A characteristic temperature can be, for example, a measured temperature, an estimated temperature, or an average of several measured or estimated temperatures for one or more points within the LNT 207 or the exhaust within or immediately downstream from the LNT 207. A characteristic temperature can be an actual value, an estimated value, a target value, or a blend of the foregoing. A target value is an objective value (set point) used by a control system controlling the heating of the LNT 207.
 As a practical matter, most measures of LNT temperature have a degree of variability. During desulfation, the LNT 207 will have a range of temperatures within its volume, particularly if it is being heated. Temperatures constantly vary over time due to perturbations in exhaust conditions, noise in temperature measurements, and lag in the system's response to temperature control measures. This is all in addition to the ranging of temperatures inherent if a control method in with lean-rich cycling is employed. Accordingly, even if a method purports to raise the LNT 207 to one particular desulfating temperature, it would be more accurate to say the LNT 207 is raised to within a desulfating temperature range.
 Descriptions of the present invention refer to a either a desulfating temperature, or a desulfating temperature range having an upper limit (peak or maximum). These descriptions are coextensive. A "desulfating temperature" can be understood as either a single characteristic temperature or the peak or average of a range for a characteristic temperature. A description referring to a range of desulfating temperatures is inclusive of cases that purport to hold the LNT 207 at a single desulfating temperature. The range can be considered either the single value (upper and lower limits the same), or the single value plus or minus a measure of uncertainty or variability, e.g. ±25° C.
 The invention can be employed regardless of how the LNT temperature is characterized. If the examination of any one characteristic temperature demonstrates the invention is being employed, then the invention is being employed regardless of whether that characteristic temperature is used by a controller. An increase in any one characteristic temperature or upper limit temperature will inherently result in increases to all other characteristic temperatures and upper limit temperatures.
 Aging of the LNT 207 refers to irreversible physical or chemical changes that occur over time with use. Aging causes a progressive deterioration in functionality. The affected functionality includes at least NOX uptake efficiency. Aging is not relieved by desulfation. Aging can occur through a variety of mechanisms, which may or may not be elucidated. Typically, however, an important mechanism of aging is thermal aging or sintering. At elevated temperatures, small catalyst particles gradually coalesce into larger particles, the rate depending on temperature. This coalescence results in a reduction in surface area. As catalyst surface area goes down, so does catalyst activity. Aging can also occur through essentially irreversible poisoning. For example, SOX or another compound can become bound in such a way that the poison cannot be removed by a standard desulfation process. The time over which LNT aging occurs is time in operation, especially time at desulfating temperatures. The higher the temperature, the more quickly aging is taking place. Aging will occur more quickly if high sulfur fuels, necessitating more frequent desulfations, are used.
 While aging can occur catastrophically, more typically aging is a process that occurs gradually over the lifetime of the LNT 207, including many desulfations. Typically, the LNT 207 has a service life spanning several hundred thousand kilometers of highway driving or the equivalent. The service life is typically thousands of hours over which hundreds of desulfations take place.
 Increases in maximum desulfating temperature are made over a substantial portion of the lifetime and many desulfations. While a small increase can be made with each desulfation, use of the present invention is better characterized by increases in the maximum desulfating temperature within successive periods, each period comprising many desulfations, such as 10, 25, 50, or 100 desulfations.
 The increases in maximum desulfation are spread out over a substantial portion of the lifetime of the product, e.g., from about 25% to about 75% of the lifetime. Cumulative increase over the lifetime are typically from about 30 to about 100° C., e.g., about 50° C. A typical rate of increase is on the order of about 5 to about 10° C. over periods of about 50 to 100 desulfation cycles.
 The intervals between desulfations are typically from about 10 to 100 operating hours, e.g. 30 hours. The intervals depend on the sulfur content of the fuel and generally decrease over time due to diminishing storage capacity. Increases of 5 or 10° C. in maximum desulfating temperature are typically made over periods of about 500 to 5,000 hours of operation.
 The operating lifetime can be divided into consecutive time intervals, I1, I2, . . . IN, with N≧2, each interval comprising at 1,000 operating hours and at least 10 desulfations. The intervals can be selected so that the highest of the desulfating temperatures used within each interval In is at least 5° C. greater than that of the preceding interval, In-1. The total increase from I1 to IN is at least 30° C., preferably at least 50° C. Preferable, N is at least 3 so that the increases are made through a series of stages. Still more preferably, N is at least 5 so that the increases are gradual, e.g., about 10° C. or less per interval.
 The increases in maximum desulfating temperature can be stepwise or continuous. They can result from the application of a continuous function to a quantified measure of the LNT's aging, or they can be made through a series of gated stages. Each desulfation need not use a higher maximum temperature than the proceeding one. The relevant increases are those being made between successive intervals each of which comprises many desulfation. The relevant increases concern the highest among the range of temperatures used within each interval.
 For example, several mild desulfations with reduced maximum temperature can be used between deeper desulfations at the current maximum temperature. The mild desulfation temperatures or the fraction of desulfations that are mild can be increased, like maximum desulfation time, as a first response to desulfation ineffectiveness before raising the maximum desulfation temperature.
 The timing of the increases to the maximum desulfation temperature can be predetermined or dynamically determined. Predetermined increases can be made, for example, according to the number of desulfations executed or the accumulated time spent at desulfating temperatures. If the aging of a particular LNT is predictable and well characterized, a predetermined schedule for increasing the maximum desulfating temperature over time has the advantage of simplicity and avoiding the possibility of premature increases resulting from measurement error. Optionally, a predetermined timing can be used to provide a minimum period to wait before increasing desulfating temperatures regardless of whether an increase is indicated by a dynamically assessment of whether a temperature increase is warranted.
 A dynamic method of increasing maximum desulfating temperatures comprises analyzing data to assess the current state or performance of the LNT 207. Such an analysis seeks to determine whether the LNT 207 can be adequately desulfated without further increases to the maximum desulfation temperature. The function of desulfation is to restore the LNT 207 to adequate levels of performance. When current desulfation condition no longer restore the LNT 207 to adequate levels of performance, an increase in the maximum desulfating temperature may be indicated. Thus, a dynamic method analyzes performance, or a measure of state that relates to performance, in order to determine whether there is a need to raise the maximum desulfation temperature.
 The data to be examined relates to the performance of the LNT 207 in terms of trapping NOX during lean phases and reducing NOX during rich phases. Measures that can be used include, for example and without limitation, NOX trapping efficiency, NOX storage capacity, frequency with which denitration is required, reductant usage during denitration, and oxygen storage capacity of the LNT 207. Any suitable data can be used. The data actually used may depend on what is available to the controller 210 as a result of the design choices made for scheduling and controlling denitration. Typically, the data considered will span at least a plurality of desulfations in order that the effects of LNT aging can be distinguished from other sources of variability in desulfation effectiveness and subsequent LNT performance.
 Some of the forgoing examples analyze LNT efficiency data to determine whether to desulfate the LNT 207 or whether to adapt the desulfating conditions. This efficiency data can be normalized to eliminate factors other than catalyst aging and sulfur loading that affect LNT performance. Such factors include LNT temperature and the exhaust flow rate. It can, however, be difficult to make data collected under disparate operating conditions comparable.
 A method for analyzing LNT performance, which is applicable to a variety of methods for determining whether to desulfate an LNT or adapt LNT desulfation conditions, sorts LNT performance data into separate groups according to the conditions under which the data was collected. Each group corresponds to a distinct range of conditions. For example, there may be 16 groups indexed according to two parameters. The two parameters can be, for example, engine out exhaust gas temperature and engine out oxygen concentration. The span of exhaust temperatures is divided into four ranges, the span of exhaust oxygen concentrations is divided into four ranges, whereby the possible combination create the 16 distinct groups.
 When a decision is to be made based on LNT performance, the decision criteria is analyzed separately for each group's data. If the criteria is met based on the data from any one group for which a statistically significant sample has been collected, the decision, such as a decision to desulfate or adapt desulfating conditions, can be made. Variations on the method include using only the group with the most recently obtained statistically significant data, only the group with the most statistically significant data, or voting among the groups having data that is both sufficiently recent and statistically significant.
 The number of parameters to use, the identity of the parameters, the number of ranges to divide each parameter into, are all choices that can be flexibly made. Suitable parameters include, without limitation those that characterize the engine's operating state, such as torque, speed, and load, and those that characterize the exhaust or LNT condition, such as air-to-fuel ratio, oxygen concentration, flow rate, exhaust temperature, and LNT temperature.
 The number or parameters and their ranges are preferably defined in such a way that the majority of operating conditions fall within a relatively small or moderate number of groups, such as 4 to 20. The number of groups is preferably limited to ensure that at least one group generally accumulates a statistically significant set of data in a timely fashion for making the relevant decision.
 Several sets of data can be retained. For example there may be one set corresponding to LNT performance within limited intervals that follow successfully completed desulfations. That data set is used for adapting desulfating conditions. There may be another data set that contains the most recent LNT performance data. That data, optionally in combination with the post desulfation data, can be used to decide whether it is time for desulfation.
 Within each group, only a limited number of data points corresponding to the most recently collected data may be retained. For example, each group may retain a number of date points in the range from 3 to 7, for example 5. As new data is collected, the oldest is discarded. Alternatively, the number of data points retained is not limited by number, but is limited by age. In using the data, an average or median value can be taken. In one example, 5 data points are retained within each set of each group. Analysis of lean NOX trap performance uses the averages taken after discarding the highest and lowest values.
Patent applications by Christian Thomas Chimner, Royal Oaks, MI US
Patent applications by Hanlong Yang, Novi, MI US
Patent applications by James Edward Mccarthy, Jr., Canton, MI US
Patent applications by Jiyang Yan, Troy, MI US
Patent applications by Eaton Corporation
Patent applications in class Condition responsive control of heater, cooler, igniter, or fuel supply of reactor
Patent applications in all subclasses Condition responsive control of heater, cooler, igniter, or fuel supply of reactor