Research on particulate filter simulation and regeneration control strategy

Mechanical Systems and Signal Processing ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Contents lists available at ScienceDirect

Mechanical Systems and Signal Processing journal homepage: www.elsevier.com/locate/ymssp

Research on particulate filter simulation and regeneration control strategy Qu Dawei, Li Jun, Liu Yu n College of Automotive Engineering, Jilin University, Changchun 130025, China

a r t i c l e i n f o

abstract

Article history: Received 2 November 2015 Received in revised form 25 May 2016 Accepted 31 May 2016

This paper reports a DPF (Diesel Particulate Filter) collection mathematical model for a new regeneration control strategy. The new strategy is composed by main parts, such as regeneration time capturing, temperature rising strategy and regeneration control strategy. In the part of regeneration time capturing, a multi-level regeneration capturing method is put forward based on the combined effect of the PM (Particulate Matter) loading, pressure drop and fuel consumption. The temperature rising strategy proposes the global temperature for all operating conditions. The regeneration control process considers the particle loading density, temperature and oxygen respectively. Based on the analysis of the initial overheating, runaway temperature and local hot spot, the final control strategy is established. & 2016 Elsevier Ltd. All rights reserved.

Keywords: Diesel Particulate Filter Regeneration Control strategy Engine

1. Introduction Diesel Particulate Filter (DPF) is widely known as an effective exhaust-particle post-processing equipment. However, when the particles are accumulated to a certain level, exhaust resistance and the DPF pressure drop will increase, which can consequently pose a adverse impact on fuel economy and output power [1,2]. Thus, timely removal of these particles is highly desired so as to achieve DPF regeneration process. In this article, a new control strategy for regeneration process is built from DPF regeneration time, temperature rising strategy and control strategy of regeneration process [3,4]. Since the vehicle running conditions is transient and complicated, there are still multiple big difficulties for the DPF regeneration time determination [5,6]. Based on the Particulate Matter (PM) load model, the pressure drop method or fuel consumption method, traditional DPF regeneration control strategy determines the regeneration time, yet it is unable to accurately predict the amount of PM loading as well as the lack of the optimal temperature rising control strategy. Additionally, insufficient care is taken towards the runaway regeneration and hot spot phenomena, which will easily cause damage to the DPF carrier and thus reduce its service life [7,8]. Firstly, a new method called “multi-level regeneration time determination”, which couples three traditional methods including the PM loading mathematical model method, the pressure drop method and fuel consumption method, is put forward to enhance the accuracy of regeneration time control and solve these problems. Along with the appropriate temperature rising, methods are determined to achieve “two temperature” (e.g. Diesel Oxidation Catalyst (DOC) inlet temperature and DPF inlet temperature) as well as temperature rising rate requirements according to the specific engine operating conditions. Finally, the appropriate method of controlling the DPF temperature distribution is determined. The n

Corresponding author. E-mail address: [email protected] (L. Yu).

http://dx.doi.org/10.1016/j.ymssp.2016.05.039 0888-3270/& 2016 Elsevier Ltd. All rights reserved.

Please cite this article as: Q. Dawei, et al., Research on particulate filter simulation and regeneration control strategy, Mech. Syst. Signal Process. (2016), http://dx.doi.org/10.1016/j.ymssp.2016.05.039i

Q. Dawei et al. / Mechanical Systems and Signal Processing ∎ (∎∎∎∎) ∎∎∎–∎∎∎

2

Nomenclature E E0 Ecake wcake dccake w ws Q L Df k0 ksoot b fCO Cin

Collection efficiency Collection efficiency of fresh DPF Collection efficiency of cake layer Thickness of particle layer Diameter of microspheres Wall thickness Thickness of particle layer Exhaust volumetric flow rate Length of carrier's channel Diameter of the carrier Permeability coefficient of wall Permeability coefficient of particle layer Side length of outlet channel Selective coefficient of CO Inlet PM concentration for DPF

Cout KR Tin Tout Q̇ r ̇ qtrans ΔHCO ΔHCO2 A0 E0 yO2 Ap Aa Ep Ea cp, filter

ρfilter

Outlet PM concentration for DPF Reaction rate of the PM DPF inlet exhaust temperature DPF outlet exhaust temperature Energy released by the oxidation Heat released to the environment from DPF Formation enthalpy of CO Formation enthalpy of CO2 Frequency factor Chemical reaction activated energy Molar fraction of O2 in the exhaust Frequency factors of thermal reaction Frequency factors of catalytic reaction Thermal reaction's activated energy Catalytic reaction's activated energy Specific heat capacity of the carrier's material Density of the porous medium

DPF regeneration control strategy proposed in this paper can accurately predict the amount of PM load, based on which determines the appropriate regeneration timing and proper temperature mode. Further, it can achieve precise control of the DOC inlet temperature, the DPF inlet temperature and the rate of temperature rise. Compared with other DPF regeneration control strategies, the control strategy proposed in this paper have finally realized the reasonable distribution of DPF temperature and made the regeneration process faster, safer and more efficiently.

2. DPF model 2.1. PM collection model As shown in Fig. 1, the DPF carrier's wall is discretized into some slabs with the same thickness. It is assumed that every slab contains the same amount of unit collectors with the same size and collects particles by themselves. The collection efficiency E of fresh DPF is drawn from Fig. 1, which can be expressed as:

E=

min−mout m −(1−Ei ) mi m −(1−Ei )(1−Ei − 1)(1−Ei − 2 ) L (1−E1) min = in = in min min min

= 1−(1−Ei )(1−Ei − 1)(1−Ei − 2 ) L (1−E1)

(1)

Assuming that the carrier's wall is composed by an infinite number of slabs and the direction of airflow movement is the positive direction of the x-axis in the figure, then the fresh DPF collection efficiency E by integrating is:

Fig. 1. Fresh DPF carrier's wall which is discretized into a number of slabs.

Please cite this article as: Q. Dawei, et al., Research on particulate filter simulation and regeneration control strategy, Mech. Syst. Signal Process. (2016), http://dx.doi.org/10.1016/j.ymssp.2016.05.039i

Q. Dawei et al. / Mechanical Systems and Signal Processing ∎ (∎∎∎∎) ∎∎∎–∎∎∎

3

Fig. 2. DPF pressure drop distribution diagram.

E=

∫ E (i, t ) dx

(2)

During the collection of particles, the particle layer plays a role of collecting particles, and it will be gradually formed in the inlet channel wall of the carrier. The previous analysis merely considers the process before the formation of particle layer. As for the particle layer, its collection efficiency is: w ⎛ −ηc cake ⎞ Ecake (t ) = A η ⎜⎜ 1−e dccake ⎟⎟ ⎠ ⎝

(3)

Where Aη signifies adjustable parameter, and it is similar with collection efficiency in Konstandopoulos theory for fresh DPF carrier's wall; ηc denotes collection efficiency for microspheres (unit collector) in particle layer, wcake refers to the thickness of particle layer, while dccake signifies the diameter of microspheres [9–11]. 2.2. Exhaust pressure drop model As shown in Fig. 2, DPF pressure drop can be mainly divided into three categories [12], such as airflow contraction and expansion losses, frictional losses in inlet and outlet channels, wall and particle layer losses. In the study, the losses brought by the contraction and expansion of the air flow are small and negligible. The latter two pressure drops are principally researched.

ΔP =ΔPfilterwall + ΔPsootlayer + ΔPinletchannel + ΔPoutletchannel

(4)

Where a denotes side length of the inlet channel, w denotes the wall thickness, while ws signifies the thickness of particle layer. In the formula (4), the first two items follow Darcy law and they can be expressed as:

ΔPfilterwall =

⎞2 ⎛ w ⎞ μQ ⎛ a + b ⎟ + w ⎟ ⎜⎜ ⎜ ⎠ ⎝ k 0 a ⎟⎠ 2Vtrap ⎝ 2 μQ

ΔPsootlayer =

2 a+b +w 2 LπD f2 ksoot

(

)

⎛ ⎞ ⎜ a ⎟ ln ⎜ ⎟ ⎜ a−2ws ⎟ ⎝ ⎠

(5)

(6)

Where μ denotes the exhaust dynamic viscosity; Q denotes the exhaust volumetric flow rate; L denotes the length of carrier's channel; Df denotes the diameter of the carrier; k0 and ksoot are respectively permeability coefficient of wall and that of particle layer; b denotes the side length of outlet channel [13]. Frictional pressure drops in inlet and outlet channels can be shown below:

ΔPinletchannel =

ΔPoutletchannel =

⎞ ⎞2 ⎛ μQ ⎛ a + b 4FL2 ⎟ ⎜ + w ⎟ ⎜⎜ 4 ⎠ ⎝ 3 ( a−2ws ) ⎟⎠ 2Vtrap ⎝ 2 ⎞ 2 ⎛ 4FL2 ⎞ μQ ⎛ a + b + w⎟ ⎜ ⎜ ⎟ ⎠ ⎝ 3b 4 ⎠ 2Vtrap ⎝ 2

(7)

(8)

Please cite this article as: Q. Dawei, et al., Research on particulate filter simulation and regeneration control strategy, Mech. Syst. Signal Process. (2016), http://dx.doi.org/10.1016/j.ymssp.2016.05.039i

Q. Dawei et al. / Mechanical Systems and Signal Processing ∎ (∎∎∎∎) ∎∎∎–∎∎∎

4

Where F equals to 28.454, and Vtrap refers to the carrier's volume. 2.3. DPF regeneration model Based on the global chemical reaction kinetics, the amount of accumulated particles in the DPF carrier is modeled by the mass and energy balance equations for the entire carrier's volume [14–16]. It is assumed that PM can be entirely oxidized into CO2 in most of the DPF mathematical models. However, some studies show that the CO in the PM oxidation products can not be ignored. The PM, which is similar with carbon simple substance, is considered as constituting a single component in practice. Considering the fact that the PM is not completely oxidized, the PM oxidation chemical reaction equation can be expressed as:

⎛ f ⎞ C + ⎜ 1− CO ⎟ O2 → fCO CO + (1−fCO ) CO2 2⎠ ⎝

(9)

Where fCO signifies selective coefficient of CO. As shown by related studies, CO selective coefficient is limited and it is slightly affected by the temperature. Below 700 °C, fCO can be regarded as the constant. The PM oxidation process of thermal regeneration and catalytic regeneration can be described by reaction Eq. (9). Where Q denotes the volumetric flow rate of the exhaust gas; Cin denotes the inlet PM concentration for DPF; Cout denotes outlet PM concentration. Through the (Eqs. (2) and 3), fresh DPF collection efficiency is provided as E0. PM capture efficiency of the cake layer is Ecake, and overall collection efficiency E for DPF can be expressed as:

E = 1−(1−Ecake )(1−E0 )

(10)

The PM deposition rate in the DPF carrier is based on time equation, and it can be expressed as:

ṁ d = ṁ in⋅E = Q ⋅Cin⋅E

(11)

Hence, the PM mass deposited in the carrier according to the mass balance equation can be calculated:

⎡ rate of mass loading ⎤ ⎡ rate of mass oxidation⎤ ⎡ rate of mass deposition⎤ ⎢ ⎥=⎢ ⎥ ⎥+⎢ ⎦ ⎣ in the filter ⎣ in the filter ⎦ ⎣ in the filter ⎦ Considering that PM is oxidized by the first kernel response, the mass balance equation can be expressed as: (Figs. 3 and 4)

dm = − KR⋅m + ṁ d dt

(12)

Where m signifies the PM mass in the DPF carrier (wall surface and the cake layer); KR signifies reaction rate of the PM. The Fig. 5 refers to a schematic diagram of DPF internal energy balance. As shown in Fig. 5, Tin (t) represents the DPF inlet exhaust temperature; Tout (t) denotes the DPF outlet exhaust temperature; T (t) stands for the DPF carrier's temperature. It is assumed that the DPF carrier's temperature is equal to Tout (t). T1 refers to the ambient temperature, the energy released by the oxidation is Q̇ r , and the heat released to the environment ̇ . from DPF is qtrans Since the energy released by PM oxidation depends on the oxidation type (thermal oxidation and catalytic oxidation), the mass of PM oxidized in the carrier and the specific heat for reaction can be expressed as:

Q̇ r = − KR ⋅ mΔH

(13)

The specific heat for the reaction can be expressed as:

ΔH = fCO ΔHCO+(1 − fCO ) ΔHCO 2

(14)

Fig. 3. The cross-sectional diagram for DPF carrier and the cross-sectional diagram for the inlet channel.

Please cite this article as: Q. Dawei, et al., Research on particulate filter simulation and regeneration control strategy, Mech. Syst. Signal Process. (2016), http://dx.doi.org/10.1016/j.ymssp.2016.05.039i

Q. Dawei et al. / Mechanical Systems and Signal Processing ∎ (∎∎∎∎) ∎∎∎–∎∎∎

5

Fig. 4. DPF mass balance diagram.

Fig. 5. Schematic diagram of DPF internal energy balance.

Where ΔHCO and ΔHCO2 represent the formation enthalpy of CO and CO2. It is assumed that particles are oxidized at certain constant temperature (T(t) ¼const), then the Eq. (7) is simplified as a first-order ordinary differential equation. At the initial condition of t ¼0, the next m (0)¼ m0, we can obtain:

m=

⎞ ṁ d ⎛ ⎜ 1−e−KR t ⎟ + m0 e−KR t ⎠ KR ⎝

(15)

Indeed, the chemical reaction rate is not a constant, yet it is the function of the temperature and oxygen content of the exhaust. The temperature based on the chemical reaction rate is calculated by the reaction rate coefficient K, and K is given by the modified Arrhenius equation:

K = A 0 Te

⎛E ⎞ −⎜ ¯ 0 ⎟ ⎝ RT ⎠

(16)

In the formula (16), A0 signifies the frequency factor; E0 represents chemical reaction activated energy; R¯ signifies the universal gas constant. Thus, the reaction rate can be expressed as:

KR = A 0 (yo2 )nTe

⎛E ⎞ −⎜ ¯ 0 ⎟ ⎝ RT ⎠

(17)

Where yO2 denotes the molar fraction of O2 in the exhaust; n denotes the reaction order for O2; T denotes the temperature of DPF carrier and equals to DPF outlet temperature. In this model, the reaction order is 1. Thus, the reaction rates for thermal regeneration and catalytic regeneration can be expressed as:

Kthe = Ap yO2 Te

⎛E ⎞ −⎜ ¯ P ⎟ ⎝ RT ⎠ ,

Kcat = A a yO2 Te

⎛E ⎞ −⎜ a ⎟ ¯ ⎠ ⎝ RT

(18)

Where Ap and Aa represent the frequency factors of thermal reaction and that of catalytic reaction, respectively. Ep and Ea represent thermal reaction's activated energy and the catalytic reaction's activated energy, respectively. Ap and Aa refer to global constants and reflect the interaction between the catalyst and the particles according to Konstandopoulos and Kostoglou's theory [17]. They can be expressed as:

Ap = (1−β ) k 0,the, A a = βk 0,cat

(19)

Where k0, the and k0, cat refer to the intrinsic reaction constants. For the current research, β can be considered as a constant. For the catalyst-coated DPF carrier, there is not only the PM oxidation and thermal oxidation but also the catalytic oxidation in the carrier. Moreover, they both have their own activated energy and frequency factors [18,19]. Hence, the mass balance equation within the carrier can be expressed as:

dm = (−Kthe−Kcat ) m + ṁ d dt

(20)

From the Eq. (20), the mass balance equation inside the carrier must be clarified by the means of the energy balance equation. According to the Fig. 5, it can be expressed as: Please cite this article as: Q. Dawei, et al., Research on particulate filter simulation and regeneration control strategy, Mech. Syst. Signal Process. (2016), http://dx.doi.org/10.1016/j.ymssp.2016.05.039i

Q. Dawei et al. / Mechanical Systems and Signal Processing ∎ (∎∎∎∎) ∎∎∎–∎∎∎

6

⎡ change in total energy ⎤ ⎡ energy created by particulat e ⎤ ⎢ ⎥=⎢ ⎥− ⎣ of the system ⎦ ⎣ oxidation inside the trap ⎦ ⎡ enthalpy change between⎤ ⎡ heat trans ferred to⎤ ⎢ ⎥−⎢ ⎥ ⎣ inlet−outlet ⎦ ⎣ the environmen t ⎦ As described above, it's assumed that DPF carrier's temperature T equals to DPF outlet temperature Tout energy balance equation can be expressed as:

(mcp,soot + ρfilter cp,filter Vw )

(t),

dT ̇ = Q̇ r−ṁ g cp, g (T −Tm )−qtrans dt

then the

(21)

Transport property of the exhaust is determined by the exhaust components, and it denotes the function of time, temperature and the exhaust molar fraction of each component [20]. In the formula (21), cp, filter signifies the specific heat capacity of the carrier's material; ρfilter represents the density of the porous medium. Furthermore, ρfilter is given by the carrier's density and porosity:

ρfilter = ρceramicmaterial × (1−ϵ0 )

(22)

The volume Vm of carrier's wall can be obtained by subtracting volume of inlet channels and outlet channels from the carrier's volume:

Vm = Vtrap−Vchannels =

πDf 2 a2 + b2 L− Ncells L 4 2

(23)

In the Eq. (11), the heat transfer rate from carrier to environment, including thermal conduction and convection, can be expressed as:

̇ = qtrans

T −T∞ Rt

(24)

Where Rt denotes the total thermal resistance from DPF carrier to the environment, and it can be calculated as:

Rt =

ln (rins/r f ) ln (rc /rins ) 1 + + 2πkins L 2π k c L h2πrc L

(25)

Where k is the thermal conductivity between two different materials; h represents the convective heat transfer coefficient from the metal walls to the environment; The Fig. 6 shows different radius r. To sum up, the energy balance equation can be expressed as:

dT = dt ⎤⎦ ̇ + Kcat ΔHcat )−ṁ g cp, g (T −Tin )−qtrans

(mcp,soot + ρfilter cp,filter Vw ) ⎡⎣ −m (Kthe ΔHthe

(26)

With mathematical calculation software, the accumulated particle mass in the carrier can be obtained by using Runge– Kutta method for Eq. (26).

3. DPF regeneration control strategy 3.1. Regeneration time determination The regeneration time determination method based on the mathematical model relies on DPF carrier's mass and energy

Fig. 6. The schematic diagram of DPF structure.

Please cite this article as: Q. Dawei, et al., Research on particulate filter simulation and regeneration control strategy, Mech. Syst. Signal Process. (2016), http://dx.doi.org/10.1016/j.ymssp.2016.05.039i

Q. Dawei et al. / Mechanical Systems and Signal Processing ∎ (∎∎∎∎) ∎∎∎–∎∎∎

start

A

7

Calculate fuel consumption

No

Start pressure drop program

Yes

Determine particle density TYP by pressure drop pulse spectrum

M1= Mmin,TYM

max

M2=(M1,Mmax)min

PM limit value SML [Mfuelmin,Mfuelmax]

PM cumulative fuel consumption [Mfuelmin,Mfuelmax]

Mmin=TYP

Mmax=(1+x%)TYP

Determine particle density TYM by mathematical model

Determine confidence interval of particle density Mmin,Mmax

TY=M2

End

Fig. 7. The flowchart of multi-level regeneration time determination.

balance equations, with specific engine operating conditions as the initial boundary conditions and application of 4-order Runge- Kutta method to iterative calculation [21,22]. Then, the PM mass in the carrier, carrier's collection efficiency E as well as other parameters can be obtained. The loading process is separated into numerous small time intervals (Δt). Furthermore, the engine operating conditions and fuel consumption are read to calculate exhaust flow rate from which the particles concentration at this condition can be acquired in combination with PM emission universal characteristics. Then, through the utilization of the carrier's collection efficiency, the collected particles during the time internal can be calculated. Meanwhile, the PM oxidation rate is determined by the engine exhaust temperature and carrier's characteristic. During this time interval, the amount of the oxide particles is calculated. Finally, PM concentration within the carrier can be obtained by integral over time, and the collection efficiency at this time is calculated [23,24]. The determination method based on back pressure by the detection of back pressure change in the actual operation judges whether the amount of deposited particles within the carrier reaches the regeneration's requirement. For simplifying the complexity of control, the exhaust mass flow control is introduced. To be specific, it makes temperature correction to change a three-dimensional MAP (speed-load-pressure drop) into a two-dimensional one (back pressure-exhaust mass flow control) so that regeneration time judgment is simplified [25,26]. The determination method on the basis of mileage and fuel consumption is founded on the fact that the emission vehicles produce at per liter of fuel changes slightly under normal conditions. Moreover, for diesel engine, low engine exhaust temperature is difficult to reach the DPF passive regeneration's requirement. Thus, the mass range of loading particle within DPF carrier can be obtained at cost of one liter of fuel by multiple tests. Beyond that, the range provides a reference for determining the regeneration time [27]. To achieve the DPF regeneration time judgment, this paper puts forward a new method “multi-level regeneration time determination” based on the former three methods, and its flowchart is shown in Fig. 7. When the engine starts, the DPF state parameters stored in ECU are read before the last stop, and then the regeneration time determination mechanism based on mathematical model and consumption model is started simultaneously. In terms of fuel consumption method, the cumulative fuel consumption is calculated after every specific time, according to which the range [Mfuelmin, Mfuelmax] of PM accumulated in the carrier is identified to determine whether PM load limit SML belongs to the range [Mfuelmin, Mfuelmax]. If not, return A. If so, pressure drop judgment program is started to determine PM loading density TYP. According to TYP, the confidence interval [TYP, (1 þx%) TYP] of PM loading density is identified, where x is decided by the specific test conditions. Furthermore, PM loading density TYM is read from mathematical model method. If TYM∈(Mmin, Mmax), DPF carrier's particles density TY¼TYM; If TYM oMmin, TY¼ Mmin; If TYM4Mmax, the TY¼Mmax. 3.2. Exhaust temperature rising DPF regeneration temperature rising strategy chiefly considers two temperatures including DOC inlet temperature and the DPF inlet temperature. Temperature rising approaches is based on Great Wall 4D83 diesel engine whose performance Please cite this article as: Q. Dawei, et al., Research on particulate filter simulation and regeneration control strategy, Mech. Syst. Signal Process. (2016), http://dx.doi.org/10.1016/j.ymssp.2016.05.039i

Q. Dawei et al. / Mechanical Systems and Signal Processing ∎ (∎∎∎∎) ∎∎∎–∎∎∎

8

Table 1 Engine performance parameters. Engine Type

Great Wall 4D83 Diesel Engine

Cylinder number Display volume Max. power Max. torque Fuel system Intake system Exhaust post processing

4 in line 2.0 (L) 110 kW/4000 rpm 320 Nm/1800–2800 rpm Bosch IV Common Rail BorgWarner VGT Turbo DOCþ DPF

Fig. 8. DPF experiment system of 4D83 diesel engine.

Fig. 9. The divided regions for engine operating condition.

parameters are listed in Table 1 and DPF experiment system is shown in Fig. 8. Secondary post injection (Post Injection) is applied to achieve DPF inlet temperature, and the other main temperature rising methods are utilized to preheat DOC temperature. According to vehicle driving conditions, engine operating conditions are cut into different regions, in which reasonable temperature rising methods are utilized. As shown in Fig. 9, the engine operating condition can be classified into seven regions from 1 to 7, which respectively correspond to the high-speed highway conditions, the low-speed highway conditions, rapid acceleration and uphill conditions, suburban operating conditions, urban conditions, downhill conditions, and low-speed urban conditions. For high-speed and high-load areas 1 and 3, the exhaust temperature was relatively high. Thus, DOC inlet temperature can achieve oxidation of unburned HC within DOC, and DPF regeneration inlet temperature will be reached by just secondary post injection. For areas 2 and 4, slightly low DOC inlet temperature in certain conditions can be improved by fuel nearly post injection preheating DOC. Hence, the regeneration temperature requirements can be obtained by the combination with fuel nearly post injection and fuel secondary post injection. For urban condition area 5, the DPF inlet temperature requirements can be reached by using two fuel post injections, and adjusting boost pressure and the EGR rate. For Please cite this article as: Q. Dawei, et al., Research on particulate filter simulation and regeneration control strategy, Mech. Syst. Signal Process. (2016), http://dx.doi.org/10.1016/j.ymssp.2016.05.039i

Q. Dawei et al. / Mechanical Systems and Signal Processing ∎ (∎∎∎∎) ∎∎∎–∎∎∎

9

Fig. 10. DPF inlet temperature rising effect at different operating conditions. (red:600 °C; yellow:550 °C). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

low-load area 6, the throttle opening is made full use of to do intake throttling and EGR rate optimization for temperature rising. For low-speed urban condition area 7, DOC inlet temperature can be improved by adjusting the EGR rate and the fuel nearly post spray, and DPF inlet temperature will be improved by fuel secondary post injection. The Fig. 10 shows effect of different regions’ temperature rising strategies on inlet DPF temperature. As displayed in Fig. 9, different temperature rising methods in different conditions obtain good result and DPF inlet temperature can reach 600 °C under most conditions. Moreover, DPF inlet temperature for the remaining small portion of conditions can reach more than 550 °C, meeting the requirements of the DPF inlet temperature regeneration. 3.3. Regeneration process control The control for DPF regeneration process is actually for particles combustion process. The particle light-off is composed by three factors, such as fuel, temperature and oxygen, which are necessary for regeneration. To achieve control for these three factors, 3 modes of DPF aging and damage, including initial overheating, runaway regeneration and hot spot, can be researched firstly. Through the reasonable control on these 3 factors, it is possible to successfully solve three problems and achieve good control on the DPF regeneration process [28,29]. 3.3.1. Initial overheating Two main reasons that lead to DPF regeneration initial overheating include high content of soluble organic fraction (SOF) in particles, and high rising rate of DPF inlet temperature. Since the tested diesel engine is equipped with after-treatment system coupling DOC and DPF, it is possible to reduce the effect of SOF on initial overheating by the application of DOC. According to the temperature rising strategy, DPF inlet temperature and the temperature rising rate can acquire the best calibration. Hence, initial overheating can be solved by coupling DOC and suitable temperature rising strategy. 3.3.2. Runaway regeneration Runaway regeneration mainly occurs in uncontrolled DPF regeneration process. When engine suddenly drops to idle in normal regeneration process, there is a certain amount of particles unoxidized in the carrier. Indeed, some particles are in the light-off state and exhaust flow suddenly decreases, but the oxygen concentration in the exhaust increases rapidly, which makes a sharp rise in carrier's temperature. The main reasons for DPF runaway regeneration include excessive particle loading, high O2 concentration and low exhaust flow. Thus, a suitable PM loading limit must be firstly determined by analyzing the impact of particle loading on the engine power and economy, especially the temperature when engine suddenly drops to idle during regeneration. The PM loading is strongly limited by temperature, thus more times of regenerations can reduce the amount of PM load limit, which is also a challenge to fuel economy and DPF life. Beyond that, the oxygen content in the exhaust can be reduced by adjusting the EGR rate. If the exhaust flow rate is too low, it can be increased by improving the engine speed to take more heat away to prevent runaway. 3.3.3. Hot spot The reasons for hot spot include uneven particle loading and uneven air movement during the regeneration process. The main means for solving the problem is to analyze the flow uniformity and increase the uniformity of air movement as much as possible. On the one hand, it shall ensure the uniformity of the particle loading, while on the other hand, enhance Please cite this article as: Q. Dawei, et al., Research on particulate filter simulation and regeneration control strategy, Mech. Syst. Signal Process. (2016), http://dx.doi.org/10.1016/j.ymssp.2016.05.039i

10

Q. Dawei et al. / Mechanical Systems and Signal Processing ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Fig. 11. The control flowchart of regeneration process.

uniform distribution of heat during the DPF regeneration process and minimize the hot spot to the utmost extent. The flowchart of DPF regeneration control is shown in Fig. 11. After ECU determines regeneration time by particle loading density, the regeneration modular is activated. Then, the engine operating condition is read to judge whether the exhaust flow and exhaust temperature are in the range at regeneration conditions. If according with that, the DOC inlet temperature is read to determine whether it is greater than T90 (the DOC inlet temperature when HC conversion efficiency exceeds 90%); If greater, it shall proceed to the program of fuel secondary post injection; If not, it shall select the appropriate temperature rising method for engine operating conditions to make DOC inlet temperature in normal state, and then proceed to the program of fuel secondary post injection. If this condition is not suitable for regeneration, the forced regeneration inquiry is issued. If allowed, the forced regeneration is started and the engine is settled to the forced regeneration conditions, and proceeds to fuel post injection program. Firstly, the fuel injection amount and injection time are read from ECU, and fuel Please cite this article as: Q. Dawei, et al., Research on particulate filter simulation and regeneration control strategy, Mech. Syst. Signal Process. (2016), http://dx.doi.org/10.1016/j.ymssp.2016.05.039i

Q. Dawei et al. / Mechanical Systems and Signal Processing ∎ (∎∎∎∎) ∎∎∎–∎∎∎

11

Fig. 12. A complete DPF regeneration process.

injection amount is corrected by using exhaust temperature and exhaust back pressure. Then, the fuel injection begins. ECU reads the data T2 from the previous temperature sensor to judge whether T2 is greater than the limit value Tlimit. If greater, the fuel post injection rate is reduced, and program waits and returns 51. If not, it is determined whether the T2 rising rate is too fast. If fast, the fuel injection amount is reduced and program waits and returns 51. If T2 rising rate is reasonable, the data P2 is read from the DPF before pressure sensor, and P3 is read from the DPF after pressure sensor to calculate DPF pressure drop PDPF ¼P2  P3. It determines whether PDPF decreases. If PDPF is not decreased, it is decided whether T3 is greater than T2. If greater, P1 is in the fault state and the fault code returns; If not, wait and return 51; If PDPF decreases, judge whether T3 is greater than the over-temperature limit. If greater, it is determined whether PDPF sharply declines. If sharply, then stop post injection and take lowering temperature method and stop regeneration. If not, T3 is in fault state and the fault code returns; If T3 is within a reasonable range, particle burning rate is determined to calculate the amount of particle regeneration, accumulated into TYQZ. Then, it needs to analyze whether the difference between TY and TYQZ (TY TYQZ) is greater than 0. If greater, program waits and returns 51; If not, the back pressure PQ is read after regeneration to determine whether PDPF belongs to [PQ (1  x%), PQ (1 þx%)]. If not, the stop regeneration and DPF is damaged, failure code returns. If so, stop post injection, change TY, TYQZ as zero, DPF collection efficiency as initial value and stored, then the regeneration ends [30,31]. In Fig. 12, a complete DPF regeneration process using the above DPF regeneration control strategy is displayed. When the regeneration time is determined, the DOC outlet temperature which equals to DPF outlet temperature is 200 °C initially, much lower than T90. By increasing the fuel mass of the first post injection, the DPF inlet temperature rises to 400 °C and DPF regeneration process is activated at about 50th seconds. With engine stability in 1800 rpm and 19.5% of load conditions, ECU controls the secondary post fuel injection (7 mg/stroke). Meanwhile, ECU is real-time monitoring whether the DPF regeneration temperature exceeds the Tlimit (700 °C). At the 150th seconds, the DPF regeneration temperature increases and exceeds 700 °C, which will cause the DPF damage on account of overheat. Thus, ECU gradually reduces the secondary post fuel injection to 5 mg/stroke so as to decrease the DPF regeneration temperature. After about 200 s, the regeneration temperature maintains at about 580 °C. The whole DPF regeneration process lasts for 500 s, and the DPF images before and after regeneration is shown in Fig. 13. By weighting the quality of DPF before and after regeneration, more than 95% particulate has been burned, proving that the DPF control strategy achieves satisfactory regeneration level. In Fig. 14, the maximum DPF regeneration temperature, the DPF regeneration time and the relative fuel consumption during the regeneration process under different control strategies have been compared. (Control strategy 1: the strategy proposed in this paper; Control strategy 2: the constant secondary post fuel injection control strategy; Control strategy 3: the temperature closed-loop control strategy.) The contrast experiment results under three different control strategies showed that the control strategy 1 can achieve a reasonable control of temperature during DPF regeneration and ensure the safety of the DPF regeneration process. Meanwhile, compared to other DPF regeneration control strategies, control method proposed in this paper achieves a precise determination of regeneration timing and temperature control of whole working conditions, which shortens the duration of the regeneration process, and reduces the fuel consumption of regeneration process.

Fig. 13. DPF image before and after regeneration.

Please cite this article as: Q. Dawei, et al., Research on particulate filter simulation and regeneration control strategy, Mech. Syst. Signal Process. (2016), http://dx.doi.org/10.1016/j.ymssp.2016.05.039i

12

Q. Dawei et al. / Mechanical Systems and Signal Processing ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Fig. 14. Performance comparison of DPF regeneration process under different control strategies.

4. Conclusion In this work, a new method called “multi-level regeneration time determination” was put forward based upon traditional determination methods (i.e. PM load mathematical model, pressure drop method and fuel consumption method) with a detailed flowchart indicating the working mechanism of the three coupled methods, which can improve the accuracy of the determination of the regeneration time. By decreasing the boost pressure, reducing the throttle opening area, delaying the fuel main injection and increasing EGR rate, it can control the DOC inlet temperature. The target DPF inlet temperature was achieved by the fuel secondary post inject, which can establish a temperature rising strategy covering all the vehicle running conditions. Notably, the full coverage of the vehicle running condition is divided into different regimes, in which different temperature rising strategies can be applied to achieve the desired temperature for the DPF regeneration requirement. Finally, analysis towards the overheating runaway temperature and hot spot in the regeneration process was made via the particle loading density, temperature and oxygen. Moreover, corresponding regeneration control strategy was put forward with a detailed flow chart.

Acknowledgments The authors are very grateful to the financial support from National Natural Science Foundation of China (No. 51306069) & International S&T Cooperation Program of China (Cooperative Research on PM2.5 External Purification Key Technologies of Gasoline Direct Injection Engine).

References [1] Matthias Bouchez, Jean Baptiste Dementhon. Strategies for the Control of Particulate Trap Regeneration. SAE paper, 2000-01-0472. [2] D. Pinturaud, A. Charlet, C. Caillol, P. Higelin, Experimental Study of DPF Loading and Incomplete Regeneration. SAE paper, 2007-24-0094. [3] J. Yang, L. Pu, Z. Wang, Y. Zhou, X. Yan, Fault detection in a diesel engine by analysing the instantaneous angular speed, Mech. Syst. Signal Process. 15 (2001) 549–564. [4] J. Chen, Randall Robert Bond, Improved automated diagnosis of misfire in internal combustion engines based on simulation models, Mech. Syst. Signal Process. 64 (2015) 58–83. [5] S.D.L. Bensaid, D. Fino Marchisio, Numerical simulation of soot filtration and combustion within Diesel Particulate Filters, Chem. Eng. Sci. 65 (2010) 357–363. [6] N.M.M. Maia, J.M.M. Silva, E.A.M. Almas, R.P.C. Sampaio, Damage detection in structures: from mode shape to frequency response function methods, Mech. Syst. Signal Process. 17 (2003) 489–498. [7] Alessandro Cozzolini, Advanced Modeling of Diesel Particulate Filters to Predict Soot Accumulation and Pressure Drop. SAE paper, 2011-24-0187. [8] C.S. Yoon, S.H. Song, K.M. Chun. Measurement of Soot Mass and Pressure Drop Using a Single Channel DPF to Determine Soot Permeability and Density in the Wall Flow Filter. SAE paper, 2007-01-0311. [9] F. Sbrizzai, P.Soldati Faraldi, Appraisal of three dimensional numerical simulation for sub-micron particle deposition in a micro-porous ceramic filter, Chem. Eng. Sci. 60 (2015) 6551–6563. [10] R. Sharifi, R. Langari, Isolability of faults in sensor fault diagnosis, Mech. Syst. Signal Process. 25 (2011) 2733–2744. [11] R. Lòpez-Fonseca, I.Elizundia Landa, kinetic study of the combustion of porous synthetic soot, Chem. Eng. Journal. 129 (2007) 41–49. [12] Sylvain Charbonnel, Cornelius N. Opris. Fundamental Diesel Particulate Filter (DPF) Pressure Drop Model. SAE paper, 2009-01-1271. [13] Thorsten Boger, Suhao He, Thomas Collins, AchimHeibel, Douglas Beall and Christophe Remy Corning Inc. A Next Generation Cordierite Diesel Particle Filter with Significantly Reduced Pressure Drop.SAE paper, 2011-01-0813. [14] K. Hinot, H. Burtscher, A.P. Weber, G. Kasper, The effect of the contact between platinum and soot particles on the catalytic oxidation of soot deposits on a diesel particle filter, Appl. Catal. B: Environ. 71 (2007) 271–278. [15] K. Walter, Diesel engine development in view of reduced emission standards, Energy 33 (2008) 264–271. [16] Shigeki Daido, Visualization of the PM Deposition and Oxidation Behavior Inside the DPF Wall. SAE paper, 2009-01-1473. [17] Athanasios G. Konstandopoulos, Margaritis Kostoglou, Souzana Lorentzou, Wall-scale Reaction Models in Diesel Particulate Filters. SAE paper, 200701-1130. [18] S. Bensaid, C.J. Caroca, N. Russo, D. Fino, Detailed investigation of non-catalytic DPF regeneration, Can. J. Chem. Eng. 89 (2011) 401–407. [19] K.S. Martirosyan, K. Chen, D. Luss, Behavior features of soot combustion in Diesel Particulate Filter, Chem. Eng. Sci. 65 (2010) 42–46. [20] Ekathai Wirojsakunchai, Eric Schroeder, Christopher Kolodziej, David E. Foster, Detailed Diesel Exhaust Particulate Characterization and Real-Time DPF

Please cite this article as: Q. Dawei, et al., Research on particulate filter simulation and regeneration control strategy, Mech. Syst. Signal Process. (2016), http://dx.doi.org/10.1016/j.ymssp.2016.05.039i

Q. Dawei et al. / Mechanical Systems and Signal Processing ∎ (∎∎∎∎) ∎∎∎–∎∎∎

[21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31]

13

Filtration Efficiency Measurements During PM Filling Process. SAE paper, 2007-01-0320. E. Bissett, Mathematical model of the thermal regeneration of a wall-flow monolith Diesel Particulate Filter, ChemEngSci 39 (1984) 1233–1244. K. Chen, K.S. Martirosyan, D. Luss, Soot combustion dynamics in a planar Diesel Particulate Filter, Ind. Eng. Chem. Res. 48 (2009) 3323–3330. V.T. Martyn, Progress and future challenges in controlling automotive exhaust gas emissions, Appl. Catal. B: Environ. 70 (1  4) (2007) 2–15. U. Kogelschatz, B. Eliasson, M. Hirth, Ozone generation in from oxygen and air discharge physics and reaction mechanisms, Ozone Sci. Eng. 10 (1988) 367–377. M.J. Biggs, S.J. Humby, A. Buts, U. Tuzun, Explicit numerical simulation of suspension flow with deposition in porous media: influence of local flow field variation on deposition processes predicted by trajectory methods, Chem. Eng. Sci. 58 (2003) 1271–1288. K.S. Martirosyan, K. Chen, D. Luss, Behavior features of soot combustion in Diesel Particulate Filter, Chem. Eng. Sci. 65 (2010) 42–46. A.G. Merzhanov, The chemistry of self-propagating high-temperature synthesis, J. Mater. Chem. 14 (2004) 1779–1786. Rasmus Cordtz, Steady State Investigations of DPF Soot Burn Rates and DPF Modeling. SAE paper, 2011-24-0181. Zhenhua Guo, Zhaoyan Zhang, Wen Peng, 3-D Transient Elastic Thermal Stress Fieldduring Diesel Particulate Filter Regeneration. SAE paper, 2007-011141. Kihyun Baek, Hyunwoo Choi, Gisong Bae, Minsoo Woo, A Study on Regeneration Strategies of Cordierite-DPF Appliedto a Diesel Vehicle. SAE paper, 2011-28-0108. Pierre Darcy, Stéphanie Guerry, Guillaume Latouchent. Improvement of SiC DPF Control Strategiesfor Uncontrolled Regenerations with theAid of Quasi 3D DPF Model.SAE paper, 2008-01-1753.

Please cite this article as: Q. Dawei, et al., Research on particulate filter simulation and regeneration control strategy, Mech. Syst. Signal Process. (2016), http://dx.doi.org/10.1016/j.ymssp.2016.05.039i