Aiaa 2009 4539

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

7th International Energy Conversion Engineering Conference AIAA 2009-4539

2 - 5 August 2009, Denver, Colorado

Application of Similarity-Based Scaling to the Design


of an Experimental, Laboratory-Scale
Convertible Stirling – Vuilleumier Hybrid Engine

Hans-Detlev Kuehl* and Ingo Geue†


Laboratory of Thermodynamics, Dortmund University of Technology, 44221 Dortmund, Germany
Tracking No. 177185

By inserting a second regenerator in the connecting duct of a γ–Stirling engine a modified


so–called hybrid cycle is created, which features both a mechanical power generation and a
heat pump effect. This cycle can in turn be converted into a Vuilleumier heat pump cycle by
adding a fourth cylinder volume on the reverse side of the compression piston, which is thus
converted into a second displacer. So it is possible to devise a convertible machine which can
be toggled between all these operating modes depending on the current power and heat de-
mand and therefore offers remarkable opportunities as a decentralized residential energy
supply system due to its variable heat–to–power ratio. Considerable savings can thus be
achieved, particularly if a larger number of such machines are operated under central
control in order to cut peak power demands and generally optimize energy supply grids
under macro–economical aspects. In order to demonstrate the feasibility experimentally at a
reduced manufacturing and operating effort, a similarity–based scaling procedure is applied
to an original full–scale machine design, which was optimized for typical operating
conditions derived from statistical data and for minimum series production cost before. This
contribution primarily addresses the theoretical fundamentals and the limitations of this
scaling procedure as well as the options to correct any errors entailed by these. Finally, the
design of an experimental machine is presented, which can be realized and operated at a
fraction of the experimental effort required for the full–scale design and is yet expected to
yield almost identical efficiencies and heat ratios according to simulation results.

Nomenclature
Latin Symbols
A = wetted area
cM = heat capacity of a regenerator matrix
C = integral heat capacity
cp, cv = specific isobaric or isochoric heat capacity
D = diameter
dh = hydraulic diameter
h = width of piston/displacer gap
k = dimensionless constant
l = length (in the direction of flow)
n = cycle speed
p = pressure
Q = heat
R = universal gas constant
T = temperature
u = flow velocity
V = volume
x = stroke

*
PD Dr.–Ing., university lecturer, Emil–Figge–Str. 70, 44227 Dortmund, Germany, non-member

Dipl.–Ing., research assistant, Emil–Figge–Str. 70, 44227 Dortmund, Germany, non-member
1
American Institute of Aeronautics and Astronautics

Copyright © 2009 by Dortmund University of Technology. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
Dimensionless numbers
Ma = Mach number
Np = dimensionless mean pressure
Nu = Nusselt number
Pr = Prandtl number
Re = Reynolds number
Reω = Valensi number
Sg = Stirling number

Greek Symbols
α = heat transfer coefficient
ε = porosity
ζ = heat ratio
η = mechanical efficiency, dynamic viscosity
κ = heat capacity ratio
λ = thermal conductivity
Λ = flow-related dead volume
ξ = flow friction factor
ρ = density
σ = material stress
χ = scaling factor
ψ = stroke to bore ratio
ω = angular velocity

Upper Indices
* = related to reference value
 = average
^ = amplitude
˙ = time derivative

Lower Indices
allow = allowable
c = cold
D = displacer
En = enthalpy flow loss
gas = referring to the working gas
h = hot
i = index referring to a particular cycle component
ln = logarithmic mean value
M = regenerator matrix
opt = optimum
ref = reference
Sh = shuttle loss
w = warm
W = wall

I. Introduction

I N most industrialized countries a large fraction of end energy consumption is allotted to space heating and hot
water supply. In Germany, for instance, this fraction amounts to more than 40% 1, and roughly the same is true for
other central European countries. Presently, most of this heat is released by direct combustion of fossil fuels. It is
therefore evident, that considerable savings in both primary energy consumption and CO2–emissions can be
achieved here either by decentralized combined heat and power production (CHP) or by heat pumps, preferably
thermally driven ones 2, 3. However, the fixed heat–to–power ratio presently impedes an ample dissemination of
decentralized CHP, whereas heat pumps have to compete with conventional, low-cost heating systems in the market.

2
American Institute of Aeronautics and Astronautics
Regenerative cycles offer solutions for either of these options – the well–known Stirling engine currently is
under development as a decentralized CHP-system at various locations 4 - 14, and several R&D–projects 15, 16, 17 have
proven that the closely related Vuilleumier cycle is well–suited for the realization of highly efficient thermally
driven heat pumps featuring very favorable operating characteristics 18. Furthermore, it is possible to conceive a third
cycle which appears to be located in the transition zone between these two, as it features both a mechanical power
generation and a heat pump (or refrigeration) effect 19. As all three are closely related to each other, the so–called
regenerative cycles apparently offer the unique opportunity to devise a convertible machine which can be toggled
between all three operating modes described by comparatively simple means. Such a combination of a decentralized
CHP system and a heat pump (or refrigerator, respectively) can therefore be adapted to varying power and heat
(and/or refrigeration) demands, thus eliminating one of the major obstacles for a further dissemination of such
systems.
Imagining a larger number of such machines, which are operated under some kind of central control (e. g. by a
regional supplier or contractor), this adaptability may also be used to optimize electrical power supply structures on
a regional or even a supra–regional level and under macro–economic aspects (rather than just optimizing the energy
supply costs of a single home), e. g. by cutting peak demand charges and conversely avoiding unwanted electrical
power generation during off-peak time, just as a by–product when a heat demand needs to be satisfied 20.
In a current project funded by the German Research Foundation, such a convertible machine shall be realized in
an experimental, laboratory–scale size. However, to make sure that the experimental results will be significant and
meaningful for a future application, at first a theoretical study was conducted in order to determine the size and the
major design data of a machine well–suited for large–scale production and widespread operation in typical
residential areas, to work out a practically feasible design for such a machine and finally optimize it under realistic
practical constraints as derived from typical year-round consumption data for a model residential district and
corresponding statistical climate data. The main results of this study have been presented in an earlier publication 21
and will be briefly summarized below.
However, as a prototype realization of the conceived machine design appeared to be infeasible within the scope
of the current project due to limitations in time and funding as well as in laboratory and machine shop facilities, it
was reduced to a smaller, laboratory-size design by application of theoretically founded, similarity–based scaling
rules. This final design can be realized at a fraction of the constructive and machining effort which would be
required for a single prototype manufacture of the full–size design, and yet, according to the scaling rules, any
characteristic dimensionless numbers should be the same, including efficiency values, heat ratios or performance
numbers. These scaling rules leading to the final, experimental design, their limitations and the workarounds that
were found to compensate for these constitute the main focus of this contribution.

II. Thermodynamic Background


According to the well–known idealized isothermal analysis of regenerative cycles, a net heat exchange is only
possible via a cylinder volume of variable size, and the amount of heat exchanged per cycle is equal to the p,V-
diagram for this volume. Correspondingly, the mechanical net work done per cycle is obtained when the p,V–
diagram for the entire cycle volume is evaluated. In Fig. 1a this is illustrated for the γ–type Stirling engine which is
assumed to operate between a “hot” temperature Th and a “warm” temperature Tw which is assumed to be that of
the heat demand in CHP operation and is therefore distinguished from the “cold” ambient temperature level Tc .
The heat input via the hot cylinder volume VDh is expressed by a comparatively large, clockwise loop area
shaded in red, and the work done per cycle is equal to loop area obtained for the entire volume (shaded in grey).
Consequently, the difference between these areas must be equal to the green shaded, anti–clockwise loop area
representing the (negative) heat rejected at Tw . However, in γ–type Stirling engine, the corresponding “warm”
cylinder volume actually consists of two sections, namely the “warm” displacer stroke volume VDw and the stroke
volume of the compression piston VP. If these are considered separately, we find that the heat rejected via VDw must
be equal to the heat input via the hot volume VDh , as we have dVDw = -d VDh , and thus the two integrals (expressed
by the p,V–loop areas) are equal, except for the sign. On the other hand, the p,V–loop area for VP is positive,
indicating a heat input at this location. So, the heat Qw indicated in Fig. 1a actually is the difference of these two.
Of course, these two separate heat flows will never be observed in a real, non–isothermal engine, because there
is a net enthalpy transfer by a non–isothermal flow through the connecting duct and thus only the resulting net heat
flow is finally rejected by the cooler. However, if this compensating net enthalpy transfer between the two “warm”
cylinder volumes is inhibited by inserting a second regenerator in the connecting duct according to Fig. 1b, the
stroke volume VP actually becomes a heat sink which may be allowed to cool down to the “cold” temperature Tc and
finally absorb heat from the ambiance, comparable to a heat pump. This amount of heat is equal to the (blue shaded)

3
American Institute of Aeronautics and Astronautics
a) γ - Stirling cycle b) “Hybrid” cycle c) γ -Vuilleumier cycle

−W T − W Tc  Tw  Qc
η= = 1− w η= = 1 −  = η=0
Qh Th Qh Tw  Th  Qh

− Qw Tw − Qw − Qw 1 − Tc Th
ξ= = <1 ξ= =1 ξ= = >1
Qh Th Qh Qh 1 − Tc Tw

Figure 1. Schematic drawings, p, V – indicator diagrams, ideal engine efficiencies η and heat pump
performance numbers ξ for the three operating modes suggested

p,V–area for this volume and therefore to the work done per cycle. On the other hand, the heat rejected at Tw via the
remaining volume VDw is increased to the amount of heat input at Th . This is a consequence of the First Law of
Thermodynamics and furthermore, we can see in Fig. 1b that the (negative) loop area for VDw is an exact mirror
image of the positive area obtained for VDh , for we have dVDw = -dVDh , as mentioned above.
Obviously, this modified cycle features both a mechanical power generation and a heat pump (or refrigeration)
effect (and will therefore be referred to as the “hybrid” cycle in the following). For the ideal cycle, it is easy to
evaluate the mechanical efficiency η and the heat ratio ξ in comparison to the original Stirling cycle, as included in
Fig. 1. As a side effect, the pressure ratio and thus the specific power output are slightly increased.
As illustrated in Fig. 1c, the hybrid cycle is similar to the well–known γ–Vuilleumier heat pump cycle on the
other hand, as the only difference is an additional fourth cylinder volume on the opposite side of the compression
piston of the former, which is thus converted into a second displacer in Vuilleumier operation. Due to the constant
cycle volume, the ideal mechanical efficiency of the Vuilleumier cycle is zero of course, whereas the heat ratio ξ
rises up to the value obtained for a reversible thermally driven heat pump (see Fig. 1c). However, it should be noted
that the pressure amplitude of the Vuilleumier cycle is far lower than that of the Stirling and the hybrid cycle.
Apparently, it is possible to convert a Vuilleumier cycle into a hybrid cycle simply by disconnecting this
additional cylinder volume (and simultaneously connecting it to the crank case in order to avoid pressure peaks). On
the other hand, a hybrid cycle can easily be converted into a Stirling cycle by simply adapting the temperature of the
4
American Institute of Aeronautics and Astronautics
compression piston to the “warm” level Tw. The disadvantage of leaving the dead volume of the second regenerator
in place, although it is no longer needed, appears to be tolerable. Alternatively, it is also possible to disconnect the
entire “cold section” of the original Vuilleumier cycle, beginning with the second regenerator (or, in a real machine,
with the adjacent “warm” heat exchanger). However, in this case, the direction of crank rotation would have to be
reversed. As this may cause trouble in practical operation, this is only considered to be a secondary option here.

III. Conception and Design of a Convertible Machine


In any case, it appears to be possible to devise a machine which can be toggled between all three operating
modes described above by comparatively simple means. Although of course none of the ideal efficiencies and heat
ratios given in Fig. 1 can be reached in practice, such a machine still offers an additional degree of freedom with
respect to the power–to–heat ratio, if compared to conventional CHP systems.
So, it is possible to select the optimum operating mode according to the current power and heat demand of the
owner or operator, respectively, or, even better, according to the demand situation of an entire residential area, if a
larger number of such machines are operated under central control, as already described above.
In order to demonstrate the feasibility and the possible savings achievable by such machines, a constructive
design was worked out and presented in a previous publication by Geue and Kuehl 21. This design, which is
illustrated in Fig. 2, includes several new concepts developed in earlier projects in order to reduce manufacturing
costs in a potential future series production and simultaneously maintain a high thermodynamic efficiency.
The heater head, for instance, closely follows a design which had been developed in an earlier R&D project
funded by the German federal ministry of
education and research (BMBF), the aim
of which was the development of a
marketable Vuilleumier heat pump
system for residential application. It
consists of three precision steel castings
connected by two horizontal brazing
joints only, and thus is largely optimized
under manufacturing aspects. It is heated
by a low–emission matrix burner
featuring an air preheater, which can be
deactivated to realize an exceptional peak
heating power output (the so–called
“integrated bivalent operation”), and an
additional flue gas heat exchanger
recovering the remaining exhaust heat
including the heat of flue gas
condensation.
The same is true for the heat
exchangers connected to the external fluid
circuits at the warm and the cold
temperature level. These are assumed to
consist of bundles of so–called micro–
multiport aluminum extrusion profiles,
which are commonly used in vehicle air
conditioning applications.
As each of these profiles may contain
a large number of flow channels of
possibly less than a millimeter hydraulic
diameter, it is thus possible to realize
highly efficient heat exchangers at a very
low dead volume, flow resistance, and
low cost. Further details about these
design concepts can be found in the final
Figure 2. Cross section of a convertible machine for Stirling, Hybrid report on the above–mentioned project by
and Vuilleumier operation Heikrodt and Heckt 18.

5
American Institute of Aeronautics and Astronautics
Table 1: Typical operating and design data
as used for optimization purposes
Mean pressure 100 bar
Speed 1000 min-1
Swept volume (hot and cold) 295 cm3
Heater wall temperature 500 °C
warm circuit temperature 30 °C
cold circuit temperature 0 °C

Furthermore, the regenerators are assumed to


consist of fibers produced by so–called crucible
melting extraction rather than from drawn wire, a
technique developed at the Fraunhofer Institute for
Manufacturing Engineering and Applied Materials
Research 22. This allows the production of highly
efficient regenerators at comparatively low cost.
Finally, the crank mechanism follows a design
originally invented and patented by Carlsen 23 for
use in γ–type Vuilleumier heat pumps, which,
Figure 3. Performance for conditions according to Table 1 however, had to be re–designed in order to stand
as obtained by 3rd order simulation the far higher forces imposed by the compression
piston in hybrid or Stirling operation.
The size and power output of the machine were chosen with respect to the market potential of a derived series
product, particularly regarding the replacement market for heating devices in typical single family homes. The
design was then optimized for typical average operating conditions according to Table 1, using a one–dimensional
third order computer code based on Eulerian coordinates (an enhanced version of the code by Kuehl and Schulz 24).
Fig. 3 displays the performance results obtained under these conditions. It should be noted that the equality of
hot end heat input and rejected heat as well as of cold end heat input and work output in hybrid mode is actually
confirmed (except for roughly 400W reflecting the thermal losses in the cold regenerator), and that the high heat
ratio of approximately 2.0 in Vuilleumier mode is only possible because the option of self-sustaining operation
(which would be available via an appropriately enlarged cross section of the cold displacer rod) was abandoned
here. Instead, a 600W power input was accepted in this mode, as an electric motor/generator is present anyway in
this design and Vuilleumier operation most likely is the preferred choice during base load periods mainly, i. e. when
there is excess electrical energy available in the grid at low cost.
Furthermore, it should be noted that the performance results are surprisingly good for Stirling operation, too.
This is partially due to the assumption of extremely favorable heat transfer conditions in the heat exchangers made
from the above–mentioned micro–multiport extrusion profiles (e. g. a hydraulic diameter of 1.2 mm only) and also
to the low power density of the machine, the Beale number of which is 0.07 only. However, it was found that
increasing the power density at constant power output would have essentially resulted in smaller stroke volumes at a
correspondingly higher pressure amplitude only, whereas the heat exchanger dimensions would just have been
affected marginally, as these are also governed by external heat transfer limitations. As to the crank mechanism, a
slightly reduced stroke would have been outweighed by higher forces and thus the need for even larger bearings. So
it was found that no major improvement of the overall design could have been achieved by increasing the Beale
number here, particularly with respect to the anticipated cost in a potential series production.

Further details concerning both the thermodynamic layout of the machine and the optimization strategy, which
accounts for all three operating modes and also for the required hot water supply at 60°C during a fraction of the
year–round operating time, are presented and discussed in Reference 21. Furthermore (as already mentioned in the
introduction), simplified economic sample calculations based on statistical energy consumption data of a model
district comprising 300 residential units in multiple dwellings and corresponding climatic data have been performed.
The results reveal that more than 12% additional annual savings compared to a conventional, separate heat and
power supply can be achieved, if the machines (84 are needed in this case) are operated under central control, e. g.
by a contractor, rather than according to the current individual needs of the single end consumers only.
6
American Institute of Aeronautics and Astronautics
Application of Similarity Theory to Regenerative Cycles

A. General Remarks
As mentioned above, the realization of a full–size prototype engine according to Fig. 2 as a proof–of–principle is
infeasible within the current project due to limited time and funding as well as laboratory and machining facilities.
So it was decided to reduce the experimental effort by realizing an appropriately scaled “model” machine instead.
However, to make sure that the results obtained from such a model machine in terms of thermodynamic efficiencies
and heat ratios actually are transferable to the full–size machine, it is essential that similarity is maintained with
respect to the thermodynamic cycle as well as to any transport phenomena associated with it, as these determine the
major losses which in turn are the reason for non–ideal performance. These losses, which of course are accounted
for in the above–mentioned simulation program, are the following:
1) Limited heat transfer in the cylinders causing the so–called adiabatic losses
2) Limited heat transfer in the heat exchangers causing space– and time–dependent temperature differences
3) Limited heat transfer in the regenerators causing the so–called reheat loss
4) Non–isothermal gas flows due to the above limitations, causing an increased pressure amplitude and thus a
modified thermodynamic cycle as well as additional losses by non–isothermal mixing
5) Fluid flow losses in all components, causing local pressure deviations and a dissipation of mechanical work
6) Regenerator losses by a limited heat capacity of the matrix, often referred to as the “temperature swing loss”
7) Thermal regenerator losses caused by flow dispersion, i. e. deviations from ideal plug flow
8) Thermal conduction losses in the regenerator matrix
9) Thermal conduction losses in the regenerator and cylinder walls as well as via the displacer pistons
10) Shuttle and enthalpy flow losses in the displacer gaps causing energy flow down the temperature gradient
11) Non–ideal gas behavior is accounted for by inclusion of the second virial coefficient
The last–mentioned issue is not necessarily a loss. However, it is generally found that non–ideal gas behavior
does not improve the performance. Furthermore, it definitely disturbs the thermodynamic similarity, as it implicates
an additional influence of the average cycle pressure, which is difficult to account for. Fortunately, this effect is of
minor importance as long as helium is used as the working fluid, and this actually is one reason why a change of the
working fluid (which would theoretically be possible as long as the heat capacity ratio κ is maintained constant, i. e.
as long as other monoatomic gases, such as other noble gases, are used) is not considered here. Instead, non–ideal
gas behavior is the only one of the enumerated mechanisms causing deviations from the ideal thermodynamic cycle,
which is neglected in the following, i. e. the working fluid is assumed to be an ideal gas.

B. Characteristic Dimensionless Numbers


To make sure that similarity is maintained with respect to the thermodynamic cycle as well as to any of the
above–mentioned losses, any dimensionless number which is characteristic for any of these must be kept at a
constant value. Several authors have contributed to identify these; a very comprehensive treatment can be found in
Ref. 25.
First of all, similarity with respect to the thermodynamic cycle can only be expected if any characteristic ratio
affecting the results of the ideal isothermal analysis is kept constant. This primarily applies to any temperature ratio,
and the easiest way to accomplish this simply is to keep all characteristic operating temperatures at their original
values. This is also recommendable because otherwise it would not be possible to maintain a constant scaling factor
for the transport properties of the working fluid (i. e. the thermal conductivity λ and the dynamic viscosity η ) at any
temperature level, for the temperature dependence of these is nonlinear. Leaving the temperature levels unchanged
implies constant values for these properties, as their pressure dependence is negligible as long as ideal gas behavior
can be assumed. Furthermore, changing the temperature levels would be impractical as it would either result in
extremely high temperatures at the hot level or extremely low values at the cold.
Besides the temperature ratios, any volume ratio must also be kept constant, i. e. the distribution of the stroke
volume and the dead volume onto the various cycle components – including connecting ducts, to be precise – must
be the same, for otherwise the ideal pressure amplitude will usually be affected and furthermore, the mass flows at
the boundaries of the cycle components and any transport phenomena depending on these will not scale correctly.
The dimensionless number most closely related to the latter can be identified as the ratio of the dead volume Vi
of the component i (e. g. a heat exchanger or a regenerator) to the volume of working fluid ∆Vgas,i swept back and
forth through it during the cycle. It may therefore be called a “flow–related dead volume”. The reciprocal is often
referred to as the “flushing ratio”. It is obvious that these ratios are also constant as long as all volume and

7
American Institute of Aeronautics and Astronautics
temperature ratios remain unchanged. Of course, this implies constant phase angles between the moving pistons or
displacers, respectively. However, it is well–known for both the γ–Stirling and the Vuilleumier cycle that the phase
angle should be 90°, and this also applies to the hybrid cycle. So, variation of the phase angle is not a topic here.
Apart from these dimensionless numbers related to single components of the cycle (i.e. cylinder volumes, heat
exchangers, regenerators, connecting ducts, etc.) there are a few that generally refer to the entire cycle. These are

the heat capacity ratio of the gas, κ = c p cv (1),

the dimensionless cycle speed, Ma = n l ref R Tref (2),

the dimensionless mean pressure, N p = p l ref ( η ref R Tref ) (3),

and the Stirling Number, Sg = N p Ma = p ( n η ref ) (4).

Here, lref, Tref and ηref designate arbitrarily chosen reference values of length, temperature and dynamic viscosity.
Identifying the dimensionless cycle speed as a Mach number Ma follows a nomenclature introduced by Organ 25 and
does not directly reflect its original meaning as a ratio of flow velocity to speed of sound. However, it still illustrates
the fact that gas dynamic similarity inevitably requires constant flow velocities anywhere in the system, if the
working fluid and the characteristic temperature levels do not change. The Stirling number Sg most clearly reveals
that similarity with respect to any transport phenomena can only be expected, if the cycle speed n and the average
cycle pressure p are modified by the same scaling factor, regardless of any changes to the geometric dimensions.
Furthermore, it is usually easy to identify the following dimensionless numbers for any cycle component i:

The temperature ratio, Ti* = Ti Tref (5)

The dead (or stroke) volume ratio, Vi* = Vi Vref (or V̂i* = V̂i Vref ) (6)

The flow–related dead volume, Λi = Vi ∆V gas ,i (7)

The stroke–bore ratio of the cylinders, Ψ i = 2 x̂i Di (8)

u 2n li 2 l*i
A component–specific Mach number, Ma i = = = Ma (9)
κ R Ti Λi κ R Ti Λi κ Ti*

 u dh ρ   * * 
An average Reynolds number, Re i =   = 2 Ma N p  l d h  (10)
 η   Λ η* T * 
 i  i

 ω d h2 ρ   *2 
An average Valensi number, Reω ,i =   = 2π Ma N  d h  (11)
 η  p  η* T * 
 i  i

Here, an asterisk generally indicates a dimensionless ratio of the designated quantity to the corresponding
reference value. In case of a cylinder volume, the applicable hydraulic diameter usually is the cylinder diameter D.
Obviously there are a lot of interdependencies between the various numbers, so that constant values for all of
them can be assured just by keeping a subset constant, e. g. the values of Ma, Np, the temperatures and the geometric
ratios. The same is true for any further dimensionless numbers which are dependent on these, e. g. those describing
the transport processes on the side of the working fluid:
8
American Institute of Aeronautics and Astronautics
 l ρ 2
The flow friction factor ξ = ∆p  u  = f (Re, Reω ,K ) (12)
 dh 2 

α dh
The Nusselt number Nu = = f (Re, Reω , Pr,K ) (13)
λ

Keeping any temperature and geometric ratio constant, it is also clear from the ideal isothermal theory that the
indicated work per cycle (if present) is directly proportional to the mean pressure and a reference volume, i. e. the
Beale number is also constant. Similar dimensionless numbers can be formed for any of the indicated amounts of
heat illustrated by the single volume p, V–loop areas in Fig. 1. Provided that similarity is also maintained for any
loss mechanism in the real machine, it can be concluded that any work or heat exchange per cycle also is
proportional to the mean pressure and a reference volume and will therefore scale with the corresponding factors.
So, constant values of the above dimensionless numbers ensure a complete similarity of the thermodynamic
cycle which the working gas actually undergoes, including any losses caused by e. g. flow friction or limited heat
transfer. Considering Eq. (12), a constant value of ξ means that a pressure drop ∆p is linearly dependent on the
density ρ, which in turn is proportional to the cycle pressure in case of an ideal gas. Any geometric scaling factors
cancel out, and the flow velocity u must be constant to ensure a constant Mach number according to Eq. (9). So, if
the volume ratios are also constant, any work dissipated by flow friction amounts to a constant fraction of the
indicated work done per cycle (or any other amount of heat or work), and similarity is thus maintained.
Similarly, the average temperature difference required for the heat transfer in a particular component can easily
be derived from Newton’s law of cooling, if Eq. (13), (6),(2), (3) and dh = 4 V/A are inserted:

Q& n p Vref Ma N p η ref R Tref d *h2


∆T = ~ = = const . (14)
α A Nu λ A 4 Nu λ V *
dh

This means that the entire gas temperature profile with respect to both time and space remains unchanged, if the
above dimensionless numbers are kept constant.
However, there are some loss mechanisms which are also (ore exclusively) affected by the property data of solid
materials, e. g. thermal conduction losses in the cylinder or regenerator walls, which are dependent on the thermal
conductivity of these, or the temperature swing loss and the thermal conduction loss in the regenerator matrix. These
require the inclusion of some additional dimensionless numbers, which are more difficult to maintain constant:

σ allow
The related allowable material stress, σ* = (15)
p

λW
The related thermal conductivity of the wall, λ*W = (16)
λref

λM
The related thermal conductivity of the matrix, λ*M = (17)
λref

ρ M cM
The related heat capacity of the matrix, C*M = (18)
ρ cp

V free V
The porosity of the matrix, ε= = (19)
Vtotal V + VM

α λW
The Biot–Number Bi = (20)
λW n ρW cW
9
American Institute of Aeronautics and Astronautics
In the regenerator, the Biot number is of minor influence as long as the thermal penetration depth expressed by
λW n ρW cW is larger that the radius of the fibers. This is generally true for common materials and designs. In
case of heat exchanger and cylinder walls, the effect of internal wall temperature swing is included in the computer
model, however, its amplitude and thus its influence is generally found to be negligible.
A constant value of σ* ensures that strict geometric similarity is maintained with respect to the design of the
cylindrical walls of a regenerator or a cylinder, i. e. the ratio of wall thickness to internal diameter is preserved. It
can be proven that in this case the thermal conduction loss through the wall scales in the same way as the net energy
flows exchanged by the cycle, if a constant thermal conductivity λW of the wall material is assumed and Barlow’s
formula may be applied. If the mean cycle pressure is increased, it may be difficult to find a wall material offering a
sufficiently high value for σallow. However, considering that the internal cycle is not directly influenced by the actual
wall thickness but rather by the thermal conduction heat flow through its annular cross section, the same result may
also be obtained if the thermal conductivity can be reduced instead of an increased allowable stress. So, the problem
can be reduced to a single dimensionless number to be kept constant 26:

λW Tref
kW = (21)
σ allow l ref R

However, as in the particular case discussed here the intention is to lower the mean cycle pressure by appropriate
scaling rather than to increase it, the problem is less severe here, for one may simply “assume” a reduced allowable
stress of the wall material and thus choose a greater wall thickness than actually required in order to ensure a correct
scaling of the thermal conduction losses.
The heat capacity ratio CM* is dependent on the density of the working gas and thus on the mean cycle pressure,
too. On the other hand, the volume-related heat capacity of most available matrix materials is roughly in the same
range. So it may be difficult to find a suitable material in case of a considerable pressure change by scaling. Again, it
can be demonstrated by consideration of simplified second order models for the temperature swing loss that it
actually depends on one dimensionless number only 25:
ε 1
kM = (22)
1− ε Λ C*M

Apparently, a non–constant CM* might be corrected by a modified porosity ε, if the hydraulic diameter of the
matrix is maintained constant by a modified fiber diameter. However, some loss mechanisms in a regenerator do not
only depend on the hydraulic diameter but also on the fiber diameter directly, and furthermore, a modified porosity
at a constant dead volume (which is essential to keep Λ and V* constant) would result in a different cross section and
thus in a different wall diameter and wall thickness, and this in turn would result in an incorrectly scaled thermal
conduction loss in both the matrix and the wall. So, the series of similarity violations entailed by this step would be
rather complex and difficult to oversee, and it is questionable whether the term “similarity” would still be adequate.
It is therefore recommended not to change the regenerator porosity to correct a similarity violation caused by CM*.

C. Scaling Procedure Maintaining Complete Similarity for the Gas Cycle


According to Eq. (6) similarity requires constant values for any volume ratios. One obvious way to ensure this
(although not the only one – see below), is to maintain strict geometric similarity, i.e. to scale any geometric
dimension of the machine by the same scaling factor χ1, regardless whether it is oriented in the direction of flow or
perpendicular to it. So, this scaling factor has to be applied to the piston diameter and stroke as well as to any tube or
fiber diameter in a heat exchanger or regenerator, and also to the lengths of such components in the direction of
flow, and to the reference length lref, whatever this may have been chosen to be. As a consequence, any cross section
or heat transfer surface scales by χ12, and any volumes including the stroke volumes and the reference volume scale
by χ13. Furthermore, considering Eq. (2) and (3), it is evident that in this case both the speed and the mean cycle
pressure have to be scaled by χ1 -1 in order to maintain constant values of Ma and Np.
As a consequence, any power and any heat flow transferred across the system boundaries scales linearly with χ1
(as the ideal values are proportional to mean pressure, speed and volume), provided that the same scaling applies to
any internal energy flow caused by the above–mentioned loss mechanisms. This issue will be discussed below.
If χ1 is chosen greater than 1, this leads to a decrease of power density by a factor χ1-2, as the entire volume of
the machine is enlarged by χ13. On the other hand, the need for small hydraulic diameters in the heat exchangers and
10
American Institute of Aeronautics and Astronautics
the regenerators, which is typical for the design of regenerative machines featuring high power densities, is
somewhat relaxed. Furthermore, a lower pressure helps to reduce wall thicknesses and hence the above–mentioned
thermal conduction losses. The value of CM* is increased by the factor χ1, if the material of the regenerator matrix
remains unchanged, and thus the temperature swing loss is also reduced. This helps to improve the performance, and
furthermore offers new scopes for a successive re–optimization of the entire design. So, the loss of power density is
compensated by performance improvements to some extent, and the well–known fact that any design inevitably is a
compromise between these two demands is once again confirmed.
If, however, the intention is to maintain similarity as far as possible, such optimization steps must be avoided.
Instead, it is essential that any internal energy flow, which is either dissipated by friction or transferred down a
temperature gradient due to any of the above–mentioned thermal loss mechanisms, scales in exactly the same way as
the external energy flows across the cycle boundaries, i.e. linearly with χ1. As demonstrated above, it is easy to
prove this for any loss mechanism based on either heat transfer or flow friction between the gas and a solid wall. So,
this applies to the losses no. 1, 2, 3, 4, 5, 7 and 10 in the above enumeration. If the thermal conductivity of the
material remains unchanged, this is also true for the thermal conduction loss in the regenerator matrix (loss no. 8).
As explained above, similarity with respect to wall conduction losses (loss no. 9) can be maintained by a voluntary
increase of wall thicknesses in case of χ1 > 1, simply assuming an appropriately reduced allowable stress.
Even the loss mechanisms associated with piston or displacer gaps (loss no. 10) actually scale correctly, as long
as the second order models derived by Urieli and Berchowitz 27 are applicable to these. Neglecting the moderate
influence of the pressure amplitude, the sum of the shuttle and the enthalpy flow loss may be written as

π x̂ 2  Th − Tc c P T   T  
Q& Sh + Q& En = D λ ⋅ + ⋅ n ⋅ p ⋅ h ⋅ ln h  ⋅  2 − ln h   (23).
  T 
2 l  h R  Tc    c  

Here, l and h designate the length and the width of the gap, Th and Tc the temperatures at either end, x̂ and D the
stroke amplitude and diameter of the piston. Applying the above scaling rules, it is obvious that this expression
scales linearly with χ1. The optimum value for the gap width can be derived in dimensionless form as

hopt Tln*
h*opt = = (24) 25.
l ref Ma N p η Pr [2 − ln(Th Tc )]
*

Obviously this expression is constant under the above scaling conditions, and so the optimum gap width hopt
apparently scales in the same way as any other length dimension in the system, i.e. linearly with χ1.
So, the only remaining caveat in application of this scaling procedure is the adjustment of CM*, i. e. the problem
to find a matrix material featuring a volume–specific heat capacity which is by a factor χ1 lower than in the original
design. Generally, it can be concluded here, that violations of similarity only occur when the loss mechanism in
question is dependent on the property data of a solid material interacting with the gas cycle. This applies to the
material of both the regenerator matrix and the outer walls. In contrast, complete similarity can be maintained for
any loss mechanism which depends on the properties of the working gas only. This scaling procedure will therefore
be referred to as “complete similarity scaling”, although this term does not account for the deficiencies mentioned.
These caveats become even more dominant and difficult to handle when scaling by this procedure is attempted in
the opposite direction, i.e. in case of χ1 < 1. On the one hand, it is possible to increase the power density of a given
machine design in this way, but on the other hand, a couple of severe problems are encountered:
The increased pressure requires a wall material featuring a correspondingly higher strength (and this will be
difficult to find provided that the original design was already optimized in this respect). Otherwise, wall thicknesses
need to be adequately increased (or internal diameters reduced at constant thickness). This causes higher conduction
losses, and besides, additional problems due to thermal stresses may arise in the resulting thick–walled designs.
Furthermore, there will be a higher temperature swing loss in the regenerators, for it is hard to find a matrix material
featuring significantly higher volume–specific heat capacities than those commonly used.
Moreover, the need to reduce the hydraulic diameters in the heat exchangers will most likely hit the limits of
feasibility with respect to both cost and manufacturing techniques. So, apparently the scope for improvements to
regenerative machines by complete similarity scaling towards an increased power density is rather marginal.

11
American Institute of Aeronautics and Astronautics
Cross Section Scaling Procedure
Another option to maintain constant values for any volume ratio of the cycle is easily explained by imagining
two identical machines operating in the same mode next to each other. Obviously, the same overall energy flows
could be obtained if each pair of identical components would be joined and replaced by a single one with double the
cross section. So, the cross section of the cylinders and pistons would be doubled as well as that of the regenerators,
and in case of heat exchangers this would apply to the number of flow channels in parallel. On the other hand, any
length in the direction of flow (including the piston strokes) and any hydraulic diameters must be kept constant (as
well as the reference length lref), and so, referring e.g. to Eq. (9), constant flow velocities require a constant speed n
and hence a constant mean pressure, too. So, none of the above problems related to the property data of solid
materials will arise here. Instead, as long as Barlow’s formula may be applied, it is easy to prove that any thermal
conduction loss will scale by the same factor χ2 which is applied to the cross sections (i.e. 2 in the above example).
However, contrarily to complete similarity scaling, there is a problem related to the losses associated with the
cylinder volumes here, for the stroke–bore ratio according to Eq. (8) is not maintained constant. This affects the in-
cylinder heat transfer as well as the shuttle and the enthalpy flow loss (no. 10 in the above listing). Furthermore, this
fact may also cause practical problems, particularly if this scaling procedure shall be used to reduce the size of the
machine, and this is the actual intention here. As mentioned above, complete similarity scaling shall be used in order
to lower the pressure and the speed of the experimental machine and simultaneously enlarge any hydraulic structures
to simplify machining, i. e. we have χ1 > 1. As a result of this, the entire machine will grow in size, and so the
second consecutive step of cross section scaling is required to reduce this size as well as the absolute heat flows and
mechanical power down to an experimentally acceptable level and to further reduce the manufacturing effort e. g. by
a smaller number of tubes in the heat exchangers. In fact, this second scaling step will be used to reduce this number
down to 1 only in case of the heat head, which will be heated electrically for experimental simplicity reasons.
This requires a scaling factor χ2 « 1 resulting in a stroke–bore ratio Ψ » 1, which is very impractical from the
experimental point of view and difficult to realize mechanically. On the other hand, the value of Ψ is of little
influence on any of the remaining loss mechanisms mentioned above, for the thermodynamic cycle and any gas
flows through the heat exchangers and regenerators only depend on the size of the stroke volumes, but not on their
shape. So there is some scope to restore similarity by selecting a different value for Ψ, and it is suggested here to
keep Ψ constant, as this allows a correct scaling of the losses according to Eq. (23) for the following reason:
The optimum gap width h according to Eq. (24) is not affected by χ2, and the gap length l in Eq. (23), which
designates the thermally relevant length spanned by the temperature difference Th -Tc, is imposed by the surrounding
annular regenerator in the design according to Fig. 2 and is therefore maintained constant, too. In order to obtain the
correct change of stroke volume (by the factor χ2) at a constant value of Ψ now, obviously both the stroke and the
cylinder diameter have to be changed by a factor of χ21/3. So, the losses according to Eq. (23) correctly scale by χ2.
It should be noted that this is not a perfect geometric scaling though, for the width and the length of the gap
remain constant. So, there is a slight scaling error with respect to the dead volume in the gap, which, however, may
be compensated by a minor reduction of the dead volume elsewhere in or next to the cylinder volume in question.
Furthermore, the cylinder cross section is only changed by a factor of χ22/3 rather than by χ2, i. e. it is too large
by a factor of χ21/3. Now, as long as Barlow’s formula is valid, the annular cross section of the wall scales in the
same way as the internal cross section it surrounds, and so the thermal conduction loss is increased. This, however,
can be compensated by using up the surplus wall thickness introduced in the first scaling step, if χ1 · χ21/3 > 1.
As to the problem of correct scaling of the in–cylinder heat transfer, first of all it should be noted that it usually
does not affect the changes of state of the working fluid significantly. Instead, these may well be assumed to be
almost adiabatic in most cases. Furthermore, the heat transfer coefficient to the cylinder wall is very difficult to
predict. Following a common approach originally derived for Diesel engines 28, the cylinder diameter is used as the
characteristic length to define a Nusselt and a Reynolds number, and the average piston speed is used as a measure
for the flow velocity in the latter. Neglecting the influence of combustion effects finally yields

0.8
αD  n x̂ D ρ 
Nu = = k ⋅   (25).
λ  η 

If both stroke and diameter are scaled by the factor χ21/3, this results in a change of α by a factor of χ21/5, whereas
the heat transfer surface A scales by a factor of χ22/3. So, the product α A is changed by a factor of χ213/15, which is
very close to the correct linear scaling factor χ2. Most likely this error is within the range of uncertainty about α.
12
American Institute of Aeronautics and Astronautics
Application of Scaling Procedures to the Original Design
The original machine design according to Fig 1 and Table 1 was now subject to a combination of the two scaling
procedures described above. The main objectives were:

• Lowering the average cycle pressure to facilitate experimental handling and to reduce potential hazards
• Avoiding small hydraulic structures which are difficult to realize experimentally
• Reduction of tribological problems and wear by a lower speed
• Reduction of cost for peripheral devices and utilities by reduced heat flows and power to be transferred
• Reduction of size and weight of the experimental machine
• Minimizing the number of flow channels required in the heat exchangers, and thus machining effort again
• Facilitating the realization of the regenerators by increasing the required fiber diameter

Whereas the first scaling step (complete similarity scaling) primarily results in a considerably enlarged machine,
the second step (cross section scaling under the exceptional constraint of constant cylinder stroke–bore ratios) is
used to finally reduce this size down to an acceptable level. As this reduction by a scaling factor χ2 < 1 applies to the
number of flow channels in the heat exchangers, there evidently is a lower limit for this factor. In the original design
the lowest number of parallel flow channels is found in the heater head and amounts to 24. So, choosing χ2 = 1/24
will result in just one tube only in the scaled machine. This implicates a considerable simplification of the design
and simultaneously offers the opportunity of convenient electrical heating, particularly as it comes along with a
considerably reduced heating load. So it was decided to select this factor for the second scaling step.
On the other hand, it appears desirable not to enlarge any of the significant dimensions beyond those of the
original machine. This also applies to the size of the stroke volumes, which also determine the dimensions of the
crank mechanism and thus a large fraction of the machine. Thus an upper limit for the scaling factor χ1 is imposed.
As this factor is the only one which also affects the internal diameters of the heat exchanger tubes, its final value
was chosen to be χ1 = 2.61 (which is slightly below this limit and thus results in smaller cylinder volumes) in order
to obtain common, readily available values for these diameters as well as for those of the piston seals.
Table 2 illustrates the effects of the first and the second scaling step on some selected machine dimensions,
including mean pressure and speed. Obviously, the scaling procedure results in a considerably increased length of
the heat exchanger tubes at a conversely reduced number and an enlarged diameter. The same basically applies to
the regenerators. Here, the length and the fiber diameter are also increased by the factor χ1, and the cross section is
reduced. This, however, has a
Table 2 Changes to selected machine dimensions due to scaling consequence for the design of the
experimental machine in so far as it is no
Original scaling scaling longer possible to arrange the heat
design step 1 step 2 exchangers and the regenerators
annularly around the cylinders, for
Scaling factor χ - 2.61 1/24
particularly the width of the regenerator
Speed 1000 rpm 383 rpm 383 rpm matrix would be so small that noticeable
wall effects were inevitable. So, it was
Mean cycle pressure 100 bar 38.3 bar 38.3 bar decided to choose an asymmetric layout
for the experimental machine featuring
Piston diameter 90 mm 235 mm 80 mm cylindrical regenerators and shell and
Stroke 46.5 mm 121 mm 43 mm tube heat exchangers except for the heat
head, which is reduced to a single
Heater head no. of tubes 24 24 1 electrically heated tube, as mentioned
above.
Heater head tube diameter 5 mm 13 mm 13 mm This modification of course appears
Heater head tube length 450 mm 1174 mm 1174 mm to entail similarity violations with respect
to the thermal conduction losses in the
Cold HE no. of tubes 1070 1070 45 cylinder and regenerator walls. However,
it is easy to prove that the sum of the
Cold HE tube diameter 1.15 mm 3 mm 3 mm annular cross sections which are relevant
Cold HE tube length 67 mm 175 mm 175 mm for this loss is constant, regardless

13
American Institute of Aeronautics and Astronautics
whether the cross sections of the cylinder and the regenerator are surrounded by a common shell or by two separate
ones. So the overall thermal conduction loss remains unchanged as long as the temperature gradient is maintained
constant for each of the two separate shells. Whereas this is self–evident for the regenerator, this is not necessarily
true for a separate piston or displacer cylinder. Instead, the axial temperature can usually be reduced considerably
compared to the design with annular regenerators, and this in fact is a particular advantage of such designs.
Furthermore, the increased lengths of the heat exchangers and the regenerators impose the need for
comparatively long displacer domes, if the dead volumes in the connecting ducts shall be maintained at their original
values. On the other hand, similarity with respect to appendix gap losses requires an unchanged temperature gradient
in the gap anyway, and so it was decided to ensure this by provision of additional thermostat jackets around the
cylinders. Furthermore, the gap with h according to Eq. (24) is confined to the zone of the temperature gradient. For
the remaining section of the domes it is minimized by a close tolerance fit to the cylinder wall.
The design of the experimental machine obtained this way is illustrated in Fig. 4. Table 3 summarized computed
performance results (excluding gear friction) as obtained for the original and the scaled version of the machine.
Obviously, the scaling procedure resulted in a considerable reduction of all energy flows and thus the experimental
operating expenses on the one hand, whereas the efficiency values and heat ratios actually have been maintained
almost constant on the other hand, the minor deviations being due to the scaling errors mentioned above, essentially
the incorrect scaling of the in–cylinder heat transfer and the temperature swing loss in the regenerators.

Figure 4. Cross section of the experimental machine design obtained by similarity–based scaling
14
American Institute of Aeronautics and Astronautics
Table 3 Comparison of heat flows, power, efficiencies and heat ratios before and after the scaling
Original design Scaled experimental design

Stirling Hybrid Vuilleumier Stirling Hybrid Vuilleumier

Heat input 11.5 kW 12.1 kW 3.94 kW 1.28 kW 1.35 kW 0.47 kW

Rejected heat flow 7.7 kW 11.5 kW 7.55 kW 0.86 kW 1.32 kW 0.88 kW

refrigeration power - 3.1 kW 3.21 kW - 0.37 kW 0.37 kW

Mechanical power 3.8 kW 3.7 kW -0.40 kW 0.42 kW 0.4 kW -0.04 kW

Efficiency 0.330 0.306 - 0.327 0.297 -

heat ratio 0.670 0.950 1.916 0.673 0.978 1.894

To avoid the latter inconsistency, several alternative materials have been investigated. Actually, some materials
featuring a reduced volume-specific heat capacity have been found, e. g. tin or (with restrictions) aluminum alloys.
However, they either cannot be used in the hot regenerator due to their low melting points, or their thermal
conductivity is so high compared to stainless steel that a far bigger inconsistency concerning the thermal conduction
loss in the matrix would arise. Non–metallic materials such as silicon or glass have also been discussed, however,
their thermal conductivity is extremely low, and processing is very difficult due to their brittleness‡. So, it was
finally found that incorrect scaling of the temperature swing loss entails the smallest error of all options.

IV. Conclusion
The presented results demonstrate that it is possible to reduce the size and thus the effort required for the
manufacture and operation of an experimental regenerative machine including the required test–bed periphery by
similarity–based scaling. According to theoretical considerations almost any characteristic dimensionless number or
ratio can be maintained constant, including efficiency values and heat ratios, although some minor caveats have to
be mentioned – particularly with respect to the regenerator temperature swing loss and the in–cylinder heat transfer.
The experimental design further demonstrates that it is also possible to solve any constructive problems generated
by changes to the geometric dimensions of the various cycle components. A comparison of the computed
performance results for both the original and the scaled experimental design confirms that efficiencies and heat
ratios show a maximum deviation of approximately 3% only. So it can be concluded that the expected experimental
results obtained from the scaled machine will allow a meaningful performance prediction for the original design.
The next step within the current project will now be an experimental verification of the computed performance
predictions for the scaled machine design according to Fig. 4, which currently is under construction.

Acknowledgments
The authors wish to express their special gratitude to the German Research Foundation (Deutsche Forschungs-
gemeinschaft) for the financial support of this project.

References
1
R. Hirschberg, Energieeffiziente Gebäude, Verlagsgesellschaft Rudolf Müller GmbH & Co. KG, Köln, 2008.
2
C. M. Bartolini, G. Comodi, et al. , "Energy and economic analysis of stirling engine application in household heating",
Internationales Stirling Forum, Osnabrück, 2006.
3
K. Heikrodt, "Vuilleumier-Wärmepumpe für die Hausheizung und Trinkwasserbereitung", 2. Osnabrücker Energie- und
Umwelttag, Osnabrück, 1999.
4
A. Baumüller, "Kleinserienfertigung und Weiterentwicklung des Solo Stirling mikroKWK-Modul", Internationales Stirling
Forum, Osnabrück, 2004.


O. Andersen, Fraunhofer Institute for Manufacturing Engineering and Applied Materials Research, personal communication
15
American Institute of Aeronautics and Astronautics
5
D. Clucas, "Whisper Gen: Commercial manufacture", Internationales Stirling Forum, Osnabrück, 2004.
6
G. Mitchell, "New energy new business; Microgen the next generation in small scale distributed power generation & home
energy", Internationales Stirling Forum, Osnabrück, 2006.
7
F. Schlappa, "Sunmachine Holz-Pellet", Internationales Stirling Forum, Osnabrück, 2006.
8
F. Steimle, J. Lamprichs, et al. , Stirling-Maschinen-Technik, C.F. Müller Verlag, Hüthig GmbH & Co. KG, Heidelberg, 2007.
9
J. Dertenkoetter, "Market Introduction of Stirling Micro CHP", Internationales Stirling Forum, Osnabrück, 2008.
10
B. Thomas, Mini-Blockheizkraftwerke, Vogel Industrie Medien GmbH & Co. KG, Würzburg, 2007.
11
A. Schlegel, J. Ganz, et al. , "Freikolben-Stirlingmaschine als Mikro-BHKW in Einfamilienhäusern", Internationales Stirling
Forum, Osnabrück, 2006.
12
A. Baumüller and S. Luft, "Serial Production of Stirling Engines, State of the Art and Difficulties", Internationales Stirling
Forum, Osnabrück, 2006.
13
D. C. G. Veitch and K. Mahkamov, "Experimental Evaluation of Performance of WhisperGen MK III Micro CHP unit",
Internationales Stirling Forum, Osnabrück, 2006.
14
V. Brignoli, "One year operation of a SOLO 161 Stirling Solar Unit in Italy", Proc. 11th ISEC, Rome, Italy, 2003.
15
L. Larsson, "Stirling Engine Development at Kockums", 12th International Stirling Engine Conference and Technology
Exhibition, Durham, 2005.
16
J. Aranceta, A. Lopez, et al. , "The long road from laboratory to market with stirling microgeneration systems." Internationales
Stirling Forum, Osnabrück, 2008.
17
J. P. Budliger, M. Lindegger, et al. , "Prototype Stirling engine comprising one single free piston and a resonance tube",
Internationales Stirling Forum, Osnabrück, 2008.
18
K. Heikrodt, R. Heckt, et al. , "Gasbetriebene Wärmepumpe zur monovalenten Raumbeheizung und Trinkwassererwärmung",
BVE Thermolift GbR, 1999.
19
H.-D. Kühl, "Ein neuartiger, bedarfsgerecht umschaltbarer Gaskreisprozess für eine integrierte, dezentrale Wärme-Kraft-Kälte-
Kopplung", Fortschrittliche Energiewandlung und -anwendung, VDI-Berichte 1594, 2001.
20
T. Woldt and W. Fichtner, "Mikro-KWK-Anlagen im Kollektiv betreiben", BWK, Vol. 61, No. 3, 2009, pp. 46-50.
21
I. Geue and H.-D. Kühl, "Design of a Convertible Stirling – Vuilleumier Hybrid System for Demand-oriented Decentralized
Cogeneration and Heat Pump Application", Proceedings International Stirling Forum, Osnabrück, 2008.
22
F. Lehnert, G. Lotze, et al. , "Herstellung, Eigenschaften und Anwendung metallischer Kurzfasern", Materialwissenschaft und
Werkstofftechnik, Vol. 22, No. 9, 1991, pp. 355 - 358.
23
H. Carlsen, Europäisches Patent, Mechanism for transferring movements between first and second linearly displaceable
bodies., Docket No. 84900502.0, 1984.
24
H.-D. Kühl and S. Schulz, "Measured Performance of an Experimental Vuilleumier Heat Pump in Comparison to 3rd Order
Theory", Proc. 25th IECEC, Reno, Nevada, 1990.
25
A. J. Organ, Thermodynamics and Gas Dynamics of the Stirling Cycle machine, Cambridge University Press, 1992.
26
H.-D. Kühl, "Wärmetransformationsprozesse ohne Phasenumwandlung - thermodynamische Konzeption, Auslegung und
Optimierung an Beispielen aus der Wärmepumpen- und Kältetechnik", Universität Dortmund, 2003.
27
I. Urieli and D. M. Berchowitz, Stirling Cycle Engine Analysis, Adam Hilger Ltd., Bristol, 1984.
28
G. Woschni, "Die Berechnung der Wandverluste und der thermischen Belastung der Bauteile von Dieselmotoren", MTZ
Motortechnische Zeitschrift, Vol. 31, No. 12, 1970, pp. 491-499.

16
American Institute of Aeronautics and Astronautics

You might also like