A LOW PRESSURE RATIO MICRO GAS TURBINE FOR DOMESTIC

advertisement
A LOW PRESSURE RATIO MICRO GAS TURBINE FOR DOMESTIC
COMBINED HEAT AND POWER
Alister Clay*, G.D Tansley
Mechanical Engineering & Design, Aston University, Birmingham, UK
*claya@aston.ac.uk
Abstract: This paper presents work on an alternative approach in delivering a 1 kW micro gas turbine (MGT) for
domestic combined heat and power (DCHP). A method of reducing stress in a centrifugal compressor impeller is
presented followed by prototype manufacture for test by direct metal laser sintering (DMLS). Low pressure ratio is
assessed on a system perspective with regard to a novel, coiled pipe-in-pipe recuperator. The pressure ratio must be
sufficiently high if better performance is to be achieved. Using a lower specific speed can significantly reduce shaft
speed if a higher dynamic pressure component is accepted.
Keywords: micro, gas, turbine, centrifugal, impeller, stress, CFD, FEA, recuperator, DMLS
MOTIVATION
The motivation for a low pressure ratio system is
described herein: Direct: Higher component efficiency
[1] Broader operating range [2]. Lower speed: Low
risk commercial ‘off-the-shelf’ bearing solution.
Improved non contact bearing stability [3] Larger size:
Reduced aerodynamic losses, less aggressive loading.
Reduce tip clearance losses. Better manufacturing
tolerances. Low specific speed: Use kinetic energy
from low diffusion ratio to stabilize flow.
INTRODUCTION
Although favorable in terms of life and
performance, air bearings remain unproven
commercially at these scales. ‘Off-the-shelf’ oil cooled
journal bearings found in turbochargers are a lower
risk commercially but with a maximum shaft speed of
approximately 220,000 rev/min [4] pressure ratio and
so engine performance are limited.
From an impeller geometry algorithm written in
Matlab a simple compressor impeller CFD
optimization study [5] found two impellers with
opposing characteristics: impeller a 170,000 rev/min
15° backsweep and impeller d 200,000 rev/min 54°
backsweep but similar performance: pressure ratios
between 2.1 – 2.2, isentropic efficiencies between 71%
- 72%. Both impellers were characterized by low
diffusion ratios: 1.13 and 0.88 and low specific speed:
0.41 and 0.47 below the 0.6 – 0.8 range proposed by
[6].
The CFD study assessed the impact of blade
backsweep and rotational speed to produce a pressure
ratio of 2.15 with a fixed impeller diameter of 40mm
for retrofit on to the Garrett GT1241 bearing platform.
Lower rotational speeds were achieved by reducing
blade backsweep and blade height. An off-design
study demonstrated a wide operating range and regular
flow, both features thought to be the result of a low
diffusion ratio and pressure ratio.
STRESS
Problem
To accommodate stress during the CFD study, the
discharge tangential velocity component, U 2 , was
maintained below 470 m/s to permit the use of AlC355
T-6 alloy [7] a common cast material for centrifugal
compressor impellers found on turbochargers. To retro
fit the new impellers on the existing bearing platform
the back face profile and bore diameter were measured
from the GT1241 compressor impeller and modeled on
the new impellers a and d.
The stress condition was assessed using a 1° hub
segment of impeller d. FEA conducted at 200,000
rev/min found a stress concentration around the bore
edge on the back face. Although design similitude was
carried over, the von Misses stress was greater than the
material strength, suggesting failure. This prompted an
investigation into the geometric condition of the
impeller.
Material
Before investigating the back face geometry, a
study into the most common compressor materials [7]
was carried out, see Table 1. The maximum stress
exceeded both the yield and tensile stress regardless of
the material used.
Whilst bearing tests up to 220,000 rev/min were
confirmed, it’s possible the smallest production
turbochargers rarely exceeded rotational speeds
beyond 160,000 rev/min during application. Further
FEA conducted in this investigation suggested
maximum speed to be around 150,000 rev/min using
AlC356 T-6 material.
Table 1: Maximum von Misses stress results at bore
edge on back face.
Material
(17-4PH)
ρ(kg/m3)
σt(N/m2)
σy(N/m2)
σMax
7833
1.207e+09
1.137e+09
2.100e+09
(AlC356
T-6)
2680
2.280e+08
1.850e+08
8.836e+08
(AlC2618
T-61)
2760
4.410e+08
3.720e+08
9.100e+08
(Ti 6Al4V)
4429
8.274e+08
1.050e+09
1.379e+09
Back face geometry
With a fatigue limit, 17-4PH stainless steel was
selected as the new material. The prospect of a
continuous running operating strategy and numerous
start stop cycles, suggested aluminum alloys without a
fatigue limit would unsuitable. Cast metals with higher
melting temperatures such as stainless steel are also
favored due to less shrinkage. Certain geometric
restrictions were imposed during the back face
geometric study to ensure the impeller would sit in the
machined profile on the journal bearing platform:
•
•
•
With subtle geometric alterations maximum stress
was within a material yield safety factor of 2. The
centrifugal loadings combined an angular velocity of
200,000 rev/min with a radial acceleration from 0 to
200,000 rev/min in 1 sec.
Maximum resultant displacement was 0.036 mm
using stainless steel at 200,000 rev/min, well within
the 0.3 mm tip clearance used for the CFD study.
0.01987 mm
A flat radial surface to allow contact with the
thrust bearing.
A tangential arc from the impeller tip to the flat
surface of the thrust bearing face.
0.644 mm rotor tip width, approximately the same
as the GT1241 compressor impeller.
Material was introduced to form a slope from the
bore edge to the thrust bearing face, see Fig. 1, to redistribute the back face mass. The stress concentration
around the bore edge was reduced by 55%. The stress
is shown to be more evenly distributed across through
the hub core and back face. The same triangulated
profile was removed from the thrust bearing by spark
erosion at HCM Engineering Ltd1 to maintain the axial
distance between the bearing impeller.
0.03086 mm
0.01016 mm
0.02280 mm
Fig. 2: Equal blade segment showing Von Misses
stress distribution with 100 times deformation.
MANUFACTURE
Fig. 1: Centrifugal stress condition of Impeller d with
triangulated thrust bearing profile.
Blade root
A 30° (360°/blade number) impeller segment was
modeled to assess the stress condition of the blade
root, see Fig. 2. Cyclic symmetry was maintained as
blade position was mirrored on each segment face. To
maintain the aerodynamic composition of the impeller
channel, geometric alterations were kept to a
minimum. For the splitters, constant radius fillets were
introduced at the root. For the blades, a variable radius
fillet was fitted wherein material along the blade root
increased with radial distance. This method maintained
the sharp leading edge whilst providing adequate stress
distribution towards impeller exit where the centrifugal
loading is largest. There were no requirements to add
rake or reduce the blade number.
1 HCM Engineering Ltd, Pedmore Road, Lye, Stourbridge, West
Midlands, DY9 7DZ. Telephone : +44 (0) 01384 422 643 | Fax:
+44 (0) 01384899210 | Email: info@hcmeng.co.uk
Impellers
The impellers were manufactured by Direct Metal
Laser Sintering (DMLS) at Materials Solutions 2 .
Materials Solutions specializes in metal powder bed
Additive Layer Manufacturing (ALM) using a fibre
laser equipped EOS M270 machine. ALM manipulates
CAD files to build 3D objects in 20 micron layers.
Metal powder is melted with the laser which builds the
part in successive layers to achieve a 100% core
density. After the DMLS process the part undergoes a
heat treatment process prior machining followed by
grit blasting. Final parts have metallurgical properties
similar to forgings and surface finish similar to
castings. Various metals are available including
stainless steel 17-4, however these prototypes were
manufactured with a high temperature nickel
superalloy C263 more common in gas turbine
combustors. It has a density of 8200 kg/m3, tensile
strength of 950 MPa and a yield strength 0.2% PS of
675 MPa. The benefits of DMLS include:
•
•
•
Efficient manufacture of working prototypes from
CAD files similar to rapid prototyping.
Delicate three dimensional thin wall manufacture
ideal for small turbomachinery components.
Overhung features.
2 Materials Solutions, Unit 8, Great Western Business Park,
McKenzie Way, Worcester, WR4 9PT. Telephone: +44 (0) 1905
732160 | Fax: +44 (0) 1905 530224 | Email:
info@materialssolutions.co.uk
•
A 25 impeller batch DMLS time would be 72
hours. Prototype process time was 8 hours + 2
days heat treatment + 5 minutes grit blasting + 5
minute machine time on the back face.
Obstacles overcome during manufacture include:
• Tapered as opposed filleted inlet blade edges.
• Inclusion of internal thread within bore.
• Skewed/overhung blades
After finding a method of preventing damage to
the thin inlet blade tips from the DMLS powder recoater, Fig. 3 (left), Material Solutions successfully
built the impeller geometry. A grit blasting method to
remove loosely adhered particles without causing
damage, Fig. 3 (right), to the thin blades was also
developed. Essentially, the micro compressors were
built without compromise and accurately replicated the
original aerodynamic and stress design intent. The
final impellers are shown in Fig. 4.
Fig. 5: Cross section of bearing assembly showing
shroud and profile of the thrust bearing.
SYSTEM
A simple coiled pipe-in-pipe recuperator micro gas
turbine system has been assessed as a potential MGT
recuperator solution using a pressure ratio of 2.15, see
Fig. 6. The recuperator seeks to utilize standard stock
materials and manufacturing techniques to provide a
simple, low risk, solution at an expected cost of 1.5
times material cost [8].
At five coil turns the recuperator is a sufficient size
to house and contain the other MGT components
within the central void as per the design intent and
produces a MGT thermal efficiency of 11.3%, below
the target value of 15%. At one coil turn, the MGT
thermal efficiency is 8%, around half the target and at
twenty coil turns the MGT thermal efficiency is 16%.
Using an optimized pipe diameter ratio the recuperator
performance is maximized within the overall volume
envelope permitted leaving component efficiency and
or pressure ratio to raise MGT thermal efficiency.
Fig. 3: Manufacturing challenges overcome.
Fig. 6: Indicative exploded view of MGT
Fig. 4: Final micro impeller pieces.
Shrouds
Each shroud was manufactured on a Fanuc
Robodrill alpha-T21iE 3 axis CNC milling machine.
An overhang volute was built as per the restrictions of
a 3 axis machine. Fig. 5 shows a sectioned view of the
bearing assembly, a drinks can is added for reference.
In MGTs, the optimum pressure ratio for
maximum thermal efficiency is much lower compared
to conventional gas turbines with higher component
efficiencies and smaller relative pressured drops [9].
The thermal/hydraulic performance of the 5 turns
recuperator was 47.63% effective and 6.33% total
system pressure drop. Based on system calculations
using the above recuperator performance, the optimum
pressure ratio is approximately 4.0 which yielded the
target thermal efficiency of 15%. The target
thermal/hydraulic performance was 75% effective and
10% total system pressure drop which would yield the
target thermal efficiency of 15% using a sub optimum
pressure ratio of 2.15.
Fig. 7 shows calculated values for impeller d if it
were designed to deliver a pressure ratio of 4.0. Due to
a low specific speed the shaft speed is still only
280,000 rev/min which although is beyond the realms
of oil cooled journal bearings, it easily achievable with
air or electromagnetic bearings. From a DCHP
perspective with requirements for low maintenance,
non contact bearings would be welcomed provided
there is a commercial advantage or design necessity.
Low specific speed impellers are thought to
compromise efficiency [10]. With the low diffusion
450
1.0
400
0.9
Total Pressure
Static Pressure
Shaft Speed
Diffusion Ratio
Specific Speed
350
300
0.8
0.7
250
0.5
200
0.4
150
Diffusion ratio & Specific speed
Pressure (kN/m2) & Shaft speed
(krev/min)
0.3
2.0
2.5
3.0
Pressure ratio
3.5
4.0
Fig. 7: Calculated values for a 40 mm diameter, 54°
backswept compressor impeller at increasing pressure
ratios.
230,000
Compressor Exit Pressure (N/m2)
220,000
210,000
Lower speed, lower
backsweep = larger
dynamic pressure
components
200,000
190,000
180,000
170,000
Small dynamic
pressure components
from lower pressure
ratios
160,000
150,000
140,000
1.85
Pressure
component/
Shaft speed
1.90
1.95
2.00
2.05
2.10
2.15
2.20
2.25
2.30
Compressor Pressure Ratio
Total 220
Total 200
Total 170
Static 220
Static 200
Static 170
Fig. 8: Pressure components of low specific speed
impellers with different backsweep and shaft speed.
16.00
40
15.00
35
System efficiency
Recuperator mass flow
Combustion inlet velocity
14.00
13.00
30
25
12.00
20
11.00
15
10.00
Recuperator mass flow (kg/s) and
combustion inlet velocity (m/s)
System efficiency (%)
characteristic which accompanies low specific speed,
raising pressure ratio will produce an ever increasing
dynamic pressure component as suggested by Fig. 8
which may also compromise combustion downstream.
To accurately assess the effect of pressure ratio on
recuperation, CFD simulations were performed on the
coiled pipe in pipe recuperator with 5 turns at
increasing pressure ratios. The boundary conditions
used included mass flows from a system analysis and
static pressure values from compressor mean line
analysis of impeller d [5], see Fig. 7. Fig. 9 shows the
effect of the different pressure ratios in terms of
recuperator and system performance.
Whilst the dynamic pressure component increases
proportionately with pressure ratio, to maintain the
same power output the increased pressure ratio is met
with a reduction in mass flow on a systems level,
which causes an overall reduction in flow speed on
entry to the combustor. Due to a large hydraulic
diameter in the recuperator, the fluid speed altogether
is relatively slow which is also beneficial for heat
exchanger effectiveness.
10
2
2.5
3
Pressure ratio
3.5
4
Fig. 9: Affect on MGT system performance using a
coiled pipe in pipe recuperator of at 5 turns.
CONCLUSION
Low pressure ratio is assessed on a system
perspective with regard to a novel, coiled pipe-in-pipe
recuperator. The pressure ratio must be sufficiently
high if better performance is to be achieved. Using a
lower specific speed can significantly reduce shaft
speed if a higher dynamic pressure component is
accepted. Using a low pressure ratio won’t eliminate
difficulties but shifts challenges to other areas.
REFERENCES
[1]
Japikse, D., Centrifugal Compressor Design and
Performance. 1996, Vermont: Concepts ETI, Inc.
[2] Came, P.M. and C.J. Robinson, Centrifugal
Compressor Design. IMECHE, 1999. 213: p.
139-155.
[3] Peirs, J., et al. Experimental verification of
compressor performance for an ultra-micro
gasturbine. PowerMEMS. 2009. Wachington DC,
USA: Imperial College London.
[4] Garrett. Garret Product Catalog, Turbochargers,
GT12 Family, GT1241-756068-1. 2005.
[5] Clay, A. and G.D. Tansley, A micro gas turbine
for UK domestic combined heat and power.
Journal of Power and Energy - Part A, 2010.
224(6): p. 839-849.
[6] Rodgers, C. Specific speed and efficiency of
centrifugal impellers. Proceedings of the 25th
IGTC. 1980. New Orleans, USA: ASME
[7] Japikse, D., Introduction to Turbomachinery. 2nd
ed. 1997, Vermont & Oxford: Concepts ETI. Inc
& Oxford Univerisity Press.
[8] McDonald, C.F., Low-cost compact primary
surface recuperator concept for microturbines.
ATE, 2000. 20(5): p. 471-497.
[9] Clay, A. and G.D. Tansley, An Analysis of Micro
Gas Turbines for UK Domestic Combined Heat
and Power. IDGTE, Power Engineer, 2010.
14(3): p. 18-34.
[10] Casey, M.V., Computational methods for
preliminary design and geometry defintion in
turbomachinery
AGARD
paper
LS-195.
Turbomachinery design using CFD, 1994. May:
p. 1-22.
Download