tmf lite process development simulia customer conference 1 tmf lite process development qingzhong...

2010 SIMULIA Customer Conference 1

TMF Lite Process Development

Qingzhong Li, Dong Wei, Bill Moser, Don Sit, Scott Thompson

Caterpillar Inc.

Abstract: A simplified manifold transient thermal analysis method has been developed and used in

engine design and validations. Traditionally full-blown manifold transient thermal analysis has

been performed using CFD and it will take weeks for completion. In this simplified method, a user

subroutine film.f is used to define the htc and gas temperature at manifold surfaces so that the

transient thermal analysis can be performed in Abaqus quickly. The method is designed to reduce

the simulation time significantly by sacrificing some accuracy. A comparison study showed that

the simplified simulation results can capture the general trend in temperature distribution and can

be used in thermal stress and fatigue analysis for quick design evaluations during the early design

stage when many design versions need to be compared in a short time as in the case for Tier 4

manifold design. The simulation time has been reduced from weeks to days. It proved to be a good

alternative to full scale CFD analysis when many design versions need to be compared at the

early design stage.

1. Introduction

TMF represents thermal mechanical fatigue analysis which is mainly used in engine components

which experience significant high temperature in service like in manifolds. An accurate TMF life

prediction is based on accurate transient temperature predictions. Currently, the best method for

predicting these temperatures is a conjugate (i.e., fluid and metal) transient CFD analysis of the

exhaust manifold system. This analysis incorporates a portion of the cylinder head, the exhaust

manifold, fasteners, and turbine housing, and it accounts for spatial variation in metal-fluid

boundary conditions as well as temporal variation through the cycle. The typical TMF thermal

cycles and standard TMF process are illustrated in Figure 1 and 2.

The drawback to an exhaust manifold CFD is that it is quite time-consuming. A model of this

complexity typically takes two to five weeks to complete. Such time duration is not quick enough

for rapid design iterations in the early phase of manifold design process. The purpose of the TMF

“Lite” process is to simplify the TMF analysis of an early manifold design such that the analysis

completion time is one week or less. A key to this compressed time schedule is bypassing the

CFD by using a simplified FE-based thermal transient analysis.

An Abaqus user subroutine film.f has been developed to define the heat transfer boundary

conditions in the center section of the manifold. It reduces the heat transfer analysis time and

overall TMF analysis time significantly. Fatigue analysis results comparison showed that the Lite

process can capture the general pattern of fatigue life in the center section and can be used in the

early design stage.

2 2012 SIMULIA Community Conference

Figure 1. Typical TMF Thermal Cycles

Figure 2. Overview of Standard TMF Process

2. Lite Model Development

The procedure was specifically developed and validated for exhaust manifold center sections of

mid-range and heavy-duty diesel engines with in-line cylinder design. As shown in Figure 3, only

center section is analyzed for thermal analysis and structural analysis instead of full CFD analysis

and full structural analysis. A transient heat transfer analysis is performed in Abaqus to replace

CFD analysis to save simulation time. An Abaqus user subroutine film.f has been developed

specifically to define the heat transfer boundary conditions for the internal ports in the center

section of the manifold.

CFD Analysis

-Create CFD model

-Run Thermal Cycle Analysis

-CFD/FEA –Post Process

FEA thermal stress Analysis

-Create FEA model

-Abaqus Assembly/ Verify Contacts

-Thermal stress analysis

Fatigue Analysis

-Convert Stress format (skin

mesh result)

-Run in-house code for TMF

- Post Processing

Center section

Time (s)t2t1

Rated gas temp

t3 t4

Idle gas temp

Gas temperature

Cylinder Head

Exhaust ManifoldTurbo-housing

(full load)

Typical Test Cycle

0

100

200

300

400

500

600

700

800

900

1000

1100

0 200 400 600 800 1000 1200

Time(Sec)

Tem

pera

ture

(K)

tc11

tc10

tc9

tc8

tc7

tc6

tc5

tc4

tc3

tc2

tc1

Temperature cycles at different locations

Center section

Time (s)t2t1

Rated gas temp

t3 t4

Idle gas temp

Gas temperature

Cylinder Head


(full load)

Typical Test Cycle

Time (s)t2t1

Rated gas temp

t3 t4

Idle gas temp

Gas temperature

Cylinder Head


(full load)

Typical Test Cycle

0

100

200

300

400

500

600

700

800

900

1000

1100

0 200 400 600 800 1000 1200

Time(Sec)

Tem

pera

ture

(K)

tc11

tc10

tc9

tc8

tc7

tc6

tc5

tc4

tc3

tc2

tc1

Temperature cycles at different locations

t4t0 t8


Figure 3. Full TMF and Lite TMF Model

2.1 Structural Boundary Conditions

Figure 4. Full TMF and Lite TMF Model

Figure 4 shows simplified structural boundary condition for the center section.

Full CFD

(Fluent)

Full

Structural

(ABAQUS)

Lite transient heat transfer

analysis model (ABAQUS)

Lite structural model (ABAQUS)

Same mesh


2.2 Thermal Boundary Conditions

2.2.1 External Thermal Boundary Conditions

The thermal boundary conditions are in the form of a heat transfer coefficient (HTC) and fluid

temperature (T). The external boundary conditions are fixed (as shown in Figure 5). The values

for the external surface, for the manifold-to-head joint surfaces and for the slip joint surface were

determined by matching results to a full TMF analysis of a manifold center section.

Figure 5. Initial Lite Model Assumption

2.2.2 Internal Thermal Boundary Conditions

The internal boundary conditions vary with time, and they are meant to be simplification of the

Diesel Engine Exhaust System Durability Test. Different methods have been proposed and

evaluated. The initial method was to assign a fixed value (an average value based on CFD

analysis) for the entire internal port. The comparison of temperature contour plots between CFD

results and initial simplified method shows that while the initial simplified method can capture the

general trend, there are obvious temperature differences at some thermal couple locations.

•Outer surface: Htc=0.05 * head-manifold: htc=1.0

•Same as internal surface/

or no BC

Internal surfaces: constant htc 0.70

*sfilm,amplitude=temp_amp_2, film amplitude=htc_int_long2

internal_long_surf_b, F, 127,0.70

Step Time (s)t2t1

Full load htc

HTC

t3 t4

Idle load htc

Abaqus usage:

Htc unit: mw/mm^2-C


In order to further improve the simplified method, the method of defining the internal htc with

Abaqus user subroutine film.f was developed. It was designed in a way that the internal passage is

divided into different regions and each region is assigned to a different htc based on the exhaust

air flow as shown in Figure 7.

Figure 6. Comparison with CFD Results

v Idle temperature contour from CFD

v The goal is to map CFD steady results

with a set of heat transfer BC

CFD

ABAQUS simplified model

0 200 400 600 800 1000

Time(Sec)

t0 t4 t8


Figure 7. Abaqus User Subroutine film.f Development

In Abaqus user subroutine film.f, the first task is to divide the internal ports into different region

automatically. This was done by asking FEA analysts to enter some coordinates at specific nodes

like in the OD of the port interface as shown in Figure 8. With these nodal coordinate inputs, the

internal passage was divided into several regions and assigned different htc values.

In order to verify whether the right htc was assigned to the right region, the htc values assigned

were written into user variables using Abaqus user subroutine UVARM and displayed in

Abaqus/Viewer. In this way it was verified that intended htc values were assigned to the right

regions.

All cylinders have

1 exhaust pulse

every 2 revs of the

engine.

This area sees 3

exhaust pulses every 2

engine revolutions.

Assume this area gets

the “high” HTC value.

This area sees 2

exhaust pulses every 2

engine revolutions.

Physics says this area

HTC “should” be ~2/3

of the “high” value.

The head ports see 1 exhaust pulse

every 2 engine revolutions. Physics

says this area HTC “should” be ~1/3

of the “high” value.

Step Time (s)t0

Full load htc

HTC

Idle load htc

t1 t2 t3 t4


Figure 8. Input for Abaqus User Subroutine film.f

Figure 9. Verification of HTC Using User Subroutine UVARM

A1B1

C1

A3

B3

C3

Head Flange Surface 3

A2

B2

C2

Abaqus usage: *sfilm

internal_long_surf, FNU, 127, 0.61

This should be the

high htc of the port


2.3 Temperature and Fatigue Results Comparison

Temperature comparison between CFD and Lite Models (fixed htc method and user subroutine

method) is shown in Figure 10. It is shown that temperature difference between CFD results and

Lite model is reduced significantly with the use of Abaqus user subroutine. The max temperature

difference at key thermocouple locations is less than 20C.

Figure 10. Temperature Comparison

Temperature Difference

between CFD and Lite Models

-20

0

20

40

60

80

TC 1

TC 3

TC8

TC11

top

edge

radi

us

Te

mp

era

ture

Dif

f (C

)

old- h=0.70

New - h=0.61

user subroutine

Temperature Comparison

between Full Model and

Lite Models

TC10TC9

TC6, 7, 8

TC4, 5

TC2

TC3

TC11

TC1

TC10TC9

TC6, 7, 8

TC4, 5

TC2

TC3

TC11

TC1

Top edge

Radius

TC3

Thermal Couple and Other

Critical Locations Used for

Comparison

TC1

TC11

TC3

TC3


Thermal fatigue analyses were performed also based on thermal stresses from both full TMF and

TMF-Lite. The fatigue life results were compared at critical locations (shown in Figure 11).

Comparison of the fatigue results is shown in Figure 12. The X axis shows the life predicted by

using full TMF while y axis represents life predicted using TMF-Lite procedure. It is shown that

most data fall in the range of one order of magnitude, and that more than 70% of data are below

Full TMF results. It also shows that the data which exceed full TMF results are less than 2x of the

CFD results. Therefore the model is generally conservative.

Figure 11. Fatigue Results Comparison Locations

L2

L3

L1

Left View

R1

R2

R3

Right View

T3

T1T2

T4

Top View

B3

B5

B4

B1

B2

Bottom View

H1 H2

Head Flange View

B6

TMF Comparison Locations


Figure 12. Fatigue Results Comparison

3. Conclusion

A. A TMF Lite Procedure has been developed and verified; Its advantages / limitations

defined;

B. Lite Process significantly reduces the TMF analysis time to 3 ~ 5 working days

(Transient FEA + Stress FEA +TMF) from about more than 3 weeks for the full TMF;

C. This process has been incorporated as an important part of the overall manifold

validation strategy;

D. TMF Lite procedure has been used in the validation of Tier 4 exhaust manifold designs.

1.E+02

1.E+03

1.E+04

1.E+05

1.E+06

1.E+07

1.E+02 1.E+03 1.E+04 1.E+05 1.E+06 1.E+07

Cycles - Full CFD

Cycle

s -

Lit

e

Lite Analysis -User subroutine

10x2x

0.5x

0.1x

tmf lite process development simulia customer conference 1 tmf lite process development qingzhong...

Documents