The document summarizes Ben Reyngoud's PhD research project on developing a novel hybrid material for high-temperature industrial pipes. The hybrid material consists of a stainless steel liner reinforced externally with tungsten wire. This is intended to improve upon the current use of nickel-based superalloys alone. The research focuses on extracting desirable properties from each component and developing a new creep testing method to evaluate the hybrid material's performance. Initial results from unreinforced and reinforced pipe rupture tests at simulated service conditions are also presented.

1. Assessing the High Temperature
Performance of a Tungsten-Stainless
Steel Hybrid Pipe
PhD Candidate: Ben Reyngoud
University of Canterbury
Supervisor: Prof. Milo Kral
Co-Supervisor: Prof. Hari Dharan (Berkeley)

2. Industry Context
• The methanol production process involves a step in which a mixture of natural gas
and steam is flowed over a nickel catalyst at 950 C within a reformer tube
• Industrial reformer tubes used in chemical plants operate continuously under
extreme conditions (950 C and internal pressure of 3.5MPa). Currently use the long-
standing paradigm for high temperature metals, nickel-based superalloys.

3. An Opportunity
• The benefits from nickel-based superalloy development is reaching a plateau – a
paradigm shift is required.
• Effectively harnessing the potential of refractory metals in creep applications could
result in:
- Improved process efficiencies
- Fewer shutdowns
- Reduced frequency of material replacement
• These improvements have significant economic implications – 250-450 tubes in a
typical plant with a total replacement cost of approx US$15M (required after approx
10 years). Sudden failure may cost US$0.35-3M per day in lost production.

4. The Project
• Develop a novel hybrid material (stainless steel reinforced with tungsten wire
external to pipe) with superior performance in high temperature/pressure industrial
applications.
• Term ‗hybrid‘ used due to purely mechanical interface between components. This
minimizes risk of diffusion and formation of brittle intermetallics.
Property 800H 316 253MA Tungsten
σUTS at room 536 515 min 600min 1510
temp (MPa)
Melting temp 1357-1385 1375-1400 1371-1432 3410-3422
( C)
Mean CTE 17.5 17.5 19.5 4.3
(10-6 m/m C)

5. The Concept
• Atomic mobility is increased at these elevated service temperatures and under
constant pressure, creep deformation is the major enemy of these pipes.
• Diametric expansion of pipe due to creep halted by refractory wire wrap. By
effectively constricting the pipe, all the creep stress is transferred from the pipe to the
wire.
• Problems
• Oxidation – probably the greatest immediate risk to the project.
• Thermal expansion mismatch – problem associated with all composite materials. Matter is
further complicated by thermal cycling in real world conditions.

6. My Focus
• Extracting the most desirable properties from each component of the hybrid –
structural strength from liner, creep strength from wire wrap and oxidation resistance
from external barrier.
• Not looking to design tungsten-based alloys. Focus on design and investigation of
oxidation barriers – at this point specifically investigating sheathing.
• Development of novel creep testing method to assess hybrid performance –
specifically barrier and reinforcement effectiveness. Mindful of keeping this creep
testing directly applicable back to industry context.

7. Rupture Tests
• Problems with conventional creep tests:
- Geometry-based – wire winding and how to reinforce in direction of applied stress.
- How to reduce sample cross section without trivialising presence of wire?
• Solution is novel creep testing method – pipe rupture tests
- Much closer to true service loading conditions
Tube furnace Solenoid valve
Relief
valve Pressure
sensor
Argon
Sample
Thermocouple cDAQ

8. Rupture Tests
• Disadvantages of test method: only gives T and p plots over time as well as overall
creep life – no information on creep rate.
• Average service conditions – 950 C and p=2.75MPa (gives σhoop=16.56MPa).
• Test conditions – for practicality, targeted 100hr creep rupture life in unreinforced
sample. Expected life determined based on experimental creep rupture time data.

9. Pressurized Pipe Tests
Approx. 1090 C
10.4MPa = (927 C) to give 100hr
1508psi (982 C)
(1038
predicted life in
C)
(1093 control pipe.
C)

10. Results – Unreinforced Control
Unreinforced Sample 1 - Entire Test Duration
1200 30
1000 25
66.5hrs
Temperature ( C)
800 20
Pressure (bar)
600 15
400 10
200 5
0 0
0 10 20 30 40 50 60 70 80 90 100
Time (hours)
Unreinforced Sample 1 - At Design Conditions
• Why isn‘t it 100 hours?
1090 30
p d mean
• Stress by - an approximation at best and complicated by air gap.
Temperature ( C)
1080 28
Pressure (bar)
mean
2t
• 1070 Geometric changes during testing – initially massive hoop stress in liner, then it 26
expands and presses hard against sheath. Both layers then begin to creep
1060
0
together. 10 20 30 40 50 60 70
Time (hours)
• Human error in reading off log scale on creep rupture plot.

11. Results – Unreinforced Control
• Max 49% increase in diameter.

12. Results – Unreinforced Control
Sheath
Liner
• Max 79% reduction in wall thickness (including collapse of air gap).

13. Results – Reinforced 1
Reinforced Sample 1 - Entire Test Duration
1200 30
1000 25
42.8hrs
Temperature ( C)
800 20
Pressure (bar)
600 15
400 10
200 5
0 0
0 10 20 30 40 50 60 70 80
Time (hours)
• Unexpected premature failure. Diagnosis: Sample 1 - At Design Conditions
Reinforced
30
•1086 Mode and region of failure resulted in collapsed, step-
29
1084
1082
like appearance inside tube.
Temperature ( C)
1080 28
Pressure (bar)
1078
• 1076
100 C cooler at collapsed point than middle of 27
furnace. 26
25
• Caused by small span of reduced section not able to
be wrapped – design flaw to blame.
0 5 10 15 20 25 30 35 40 45 50
Time (hours)
• Reduced but unreinforced section of pipe expanded
against sheath but adjacent reinforced region
remained effectively unchanged (created step).

14. Results – Reinforced 1
• Positives:
• Only 4.5% increase in diameter and 28% reduction in thickness (including air
gap collapse) at point of rupture.
• In reinforced region, no discernable change in diameter or wall thickness.
• No signs of tungsten wire oxidation.

15. Change of Design Conditions
• In second round of testing, temp. dropped to help preserve thermocouple and pressure raised
proportionally.
Approx. 1030 C to
give 100hr predicted
life in control pipe at
14.6MPa = 2115psi new pressure.
(927
10.4MPa = 1508psi C)
(982 Approx. 1090 C to
C)
(1038
C)
(1093 give 100hr predicted
C)
life in control pipe at
old pressure.
• Step-down in liner removed to have constant reduced OD and wire wrap over full length.

16. Results – Control 2
• Control pipe without step-down in liner tested at new testing conditions to confirm
validity of changes.
• Rupture recorded after 66.6 hours (compared to 66.5 hours for original control at
higher temp and lower pressure).
• Remarkably similar appearance and creep life suggests that both pipes were valid
control tests to compare reinforced pipe results against.

17. Results – Reinforced 2
Reinforced Sample 2 - Entire Test Duration
1000 40
857hrs
Temperature ( C)
Pressure (bar)
500 20
0 0
0 100 200 300 400 500 600 700 800 900
Time (hours)
• Test stopped prematurely by power outage caused by Feb 22 earthquake but
impressive life extension nonetheless – at least 12.9x extension in life compared to
control test.
• Questions raised by test: Mode of (eventual) failure? Expected life?

18. Results – Reinforced 2
• Post-earthquake recovery of sample revealed it was still intact in furnace and fully
pressurized (i.e. had not ruptured).
• Bending/twisting of the pipe was observed in multiple planes and could not simply be
explained by the pipe sagging. Lack of damage to the furnace suggested that this
bending/twisting was not earthquake-related.
• 4.55% increase in length of the pipe not including the bends, accompanied by a
maximum 5.3% reduction in diameter and 22% reduction in wall thickness

19. Results – Reinforced 2
• Interesting mode of failure in the ‗super pipe‘
• By transferring hoop stress from liner to wire, longitudinal stress is now path of
least resistance and creep in this direction dominates. ‗Snaking‘ a result of hot
spots and some localised regions creeping faster than others
• End cap welds connect liner and sheath, meaning both are subjected to
longitudinal strain (hence both creeped).

20. Results – Reinforced 2
• While longitudinal stress is significantly lower than hoops stress, in theory it would
eventually lead to the development of circumferential cracking.
→ Assuming longitudinal stress is half of predicted hoop stress, point on creep
rupture plot is shifted to give life on the order of tens of thousands of hours.
• Not feasible to test this failure mode with these samples. But question remains: at
what point does the pipe contraction render it unusable and what can be done to
mitigate this longitudinal creep?

21. Generalized Approach
• Flaw with the approach thus-far to the design problem. Focused in on specific
geometry and composition very early, severely limiting potential for optimization.
• More general, Ashby-inspired ‗materials by design‘ approach is proposed
→ Understand the physics of the system and tailor the macro/microstructure to suit.
• In theory, this will allow for potential of individual components to be maximised while
also allowing for unit weight/cost to be minimized.
• More general approach also allows for industrial applications beyond the
petrochemical industry to be considered.

22. Ashby?
• Mike Ashby and Yves Bréchet have built on Ashby‘s well-established methodology to
explore ways of designing hybrid materials to allow for a superposition of their
properties.
• Here, a hybrid material is defined as ―a combination of two or more materials in a
predetermined geometry and scale, optimally serving a specific engineering purpose‖
– paraphrased as ―A + B + shape + scale‖.
• By treating the hybrid as a whole and selecting materials based on a criterion of
excellence derived from the physics of the system (stiffness per unit mass etc),
design solutions can be reached which are optimized for a desired property.

23. ‘Ashby chart’ Example
Designing a hybrid—here, one with high strength and high electrical conductivity. The figure shows
the resistivity and reciprocal of tensile strength for 1700 metals and alloys. We seek materials with
the lowest values of both. - M.F. Ashby, Y.J.M. Bréchet / Acta Materialia 51 (2003) 5801–5821

24. ArchiMat
• The ―A + B + shape + scale‖ mindset is a good starting point, but it largely ignores a
great deal of difficulties encountered in the successful design and fabrication of a
hybrid.
• Part of the difficulty stems from the multitude of possible choices; choice of materials,
choice of processes to combine them, and choice of the internal geometry and
topology of the constitutive materials.
• The complexity associated with applying the ‗Ashby method‘ to real world industry
cases , particularly with a finite life design, is exactly why I am currently attending
ArchiMat2011. Here I will be attending seminars on materials properties and
selection, modelling, optimization, specific architectures and specific experimental
methods.
• In addition to this, Mike and Yves will be on hand along with a panel of experts to
take part in roundtable discussions of design problems and offer their insight.

25. Multi-layered Pipe
External layer
Mid layer
Internal layer
• Goal: Tailor hybrid in terms of materials selection and architecture such that it is
optimized for a given industrial application.
• Properties to consider include: creep life/rate, structural strength, thermal
conductivity, resistance to environmental attack, compatibility between layers
(reactivity, intermetallic formation, thermal expansion etc), stiffness/ability to form
curved sections, service temperature.
• Geometric degrees of freedom: layer thicknesses, overall diameters, architecture of
mid layer – features present, their size and frequency/spacing.

26. Example of possible configuration
Rings – to facilitate thermal conductivity if Wire (wrapped) – To reinforce internal layer
required. DOF – material, ring width, spacing. (particularly re: creep) and possibly to provide
Consider potential creep and/or thermal thermal insulation/some form of heat sink
expansion, possible need for flexibility, how it (particularly if there are no rings). DOF –
will influence the effectiveness of other material, wire diameter, wrap angle, wrapped
components – i.e. wires ability to take creep span length.
strain from internal layer.

27. Preliminary Work - Screening
• Screening materials candidates based on suitability for use at 1000 C
Material Tm* ( C) σy** (MPa) ρ (Mg/m3 or α (10-6/ C) λ (W/mK)
g/cm3)
W 3410-3422 690-3447 17.6-19.25 4.3-4.5 80-170
SS 1375-1450 170-1000 7.6-8.1 13-20 11-19
Ni-alloys 1435-1466 70-1100 8.83-8.95 12-13.5 67-91
Si3N4 2388-2496 524-5500 3-3.29 3.2-3.6 22-30
Al2O3 2004-2096 690-5500 3.5-3.98 7-10.9 30-38.5
SiC 2152-2500 1000-5250 3-3.21 4-5.1 115-200
B4C 2372-2507 2583-5687 2.35-2.55 3.2-3.4 40-90
AlN 2397-2507 1970-2700 3.26-3.33 4.9-6.2 80-200
Silica glass 957-1557 1100-1600 2.17-2.22 0.55-0.75 1.4-1.5
Brick 927-1227 50-140 1.9-2.1 6-13 0.46-0.73
Stone 1227-1427 34-248 2.5-3 5.8 5.4-6
* Tg (glass transition temp) for silica glass
** Compressive strength used for ceramics (~10x higher than strength in tension).
Data from M. Ashby, Materials Selection in Mechanical Design, 3rd Ed (2007).
• Other refractory metals such as Nb, Mo, Ta and Re should also be considered.

28. Preliminary Work – Thermal Resistivity
• Remembering that these tubes exist to permit an endothermic reaction and transfer
heat from the furnace to the gases inside the pipe, it makes sense to consider a
simple model of hybrid thermal conductivity.
• Based on currently used design materials and layer thicknesses. Introduce a thermal
bridge and see how overall thermal resistivity is affected by bridge surface area.
Sheath (253MA)
Q
Air
Bridge
Wire (tungsten)
Liner (253MA)
• i.e. Vary width of thermal bridges to control how much heat is transferred through the
bridges and how much goes through the ‗air + tungsten‘ section.

29. Preliminary Work – Thermal Resistivity
Effect of thermal bridge on total resistivity of pipe hybrid
2.50E-02
253MA bridge (k=29 W/mK)
2.00E-02
Thermal resistivity (K/W)
Tungsten bridge (k=113 W/mK)
1.50E-02
1.00E-02
5.00E-03
0.00E+00
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fraction of liner surface occupied by thermal bridge
• Air gap itself makes up majority of overall resistivity and any alternate path of lesser
resistance immediately has a dramatic effect. Rapidly diminishing returns.
• Thermal conductivity of bridging material has little overall effect.

30. Preliminary Work – Wall Thickness
• Shown that models with 2.5% of liner surface bridged will still reap the majority of
thermal conductivity benefits.
• Can now assess effect of changing wall thickness on overall longitudinal stress and
thermal resistivity.
Q
R
σlong
p
• Assumes: Fixed 2.5% of surface bridged by 253MA, fixed wire diameter and air gap
thickness, wire wrap offers no support in longitudinal direction, welds at end caps
mean equal force in sheath and liner.

31. Preliminary Work – Wall Thickness
Potential Current
change
• Currently have 1.49mm thick liner and 3.95mm thick sheath – based on stock pipe
sizes and machining capabilities. Risk of wire crushing liner if thinned too much, but
no such risk in sheath – so surely it can at least be thinned to 1.49mm to match liner.
• i.e. Potential for immediate change to 3mm total thickness – would result in 3% drop
in resistivity.

32. Preliminary Work – Mass and Cost
• Denoting this proposed thickness change as ‗slim‘ reinf., some mass/length and
costing analysis is then possible.
• This assumes pipes could be made to size (i.e. no machining required) and uses a
cost of 253MA of $49.13/kg (based on current Sandvik pricing).
• For the current pipe diameter, perfect wire winding (no gaps between windings) gives
242m of Ø0.49mm wire required per m of pipe.
• The mass and cost of the theoretical thermal bridges are neglected here.

33. Preliminary Work – Mass and Cost
Wire
253MA cost/pipe
Mass/length Mass w/o wire cost/length length Total cost
(kg/m) (kg/m) ($NZ/m) ($NZ/m) ($NZ/m)
Unreinf. 5.98 5.98 293.70 0 293.70
Current reinf 6.49 5.61 275.46 313.70 589.16
'Slim' reinf. 3.97 2.84 139.70 313.70 453.40
'Slim' reinf. with
alternate wire
source 3.97 2.84 139.70 98.59 238.29
Wire cost estimated as $US1.05/m of wire. Significantly cheaper tungsten sources available, but questions over quality and
manufacturer reliability. Hubei Fotma Machinery Co (China) has tungsten at $US0.33/m of wire.
• Comparing unreinf. and ‗slim‘ reinf. pipes – potential for 13x increase in life
coupled with 33.6% decrease in mass and 18.9% decrease in cost.
→ Pending test of ‗slim‘ reinf. pipe to ensure it‘s okay in longitudinal creep and
sourcing of reliable tungsten supplier.

34. Summary
• Through careful geometric application and design considerations, overall hybrid
creep strength can be controlled by creep strength of refractory reinforcement.
• By utilizing the structural and oxidation resistant properties of the sheath and liner in
conjunction with the creep properties of tungsten, an increase in creep life in excess
of 13x is achievable.
• Through further optimization, this can potentially be achieved without any increase in
cost or mass.

35. Future Work
• From here:
• Continued testing – repeat control tests and assess creep life of ‗slim‘
reinforced pipe. Iterative process with continued design improvements.
• Design – Optimization of general pipe hybrid in terms of material
selection, geometry and architecture of mid layer.
• Modelling – combined effects of creep and thermodynamic stresses, heat
transfer for process efficiency – how thermal bridges will affect creep
response. Effect of changing wrap angle. Extension of models to consider
effects of thermal cycling.