OF
PARABOLIC TROUGH SOLAR
COLLECTORS FOR TECHNICALLY
LESS DEVELOPED COUNTRIES
TECHNICAL REPORT 1
by
Halil M. Gaven
Richard B. Bannerot
Mechanical Engineering Depariment
University o! Houston  University Park
Houston, Texas 77004
June, 1984
Any opinions, findings conclusions or recommendations
expressed in this publication are those of the author(s)
and do not necessarily reflect the views of the U.S.
Agency for International Development (USAID).
OPTICAL AND THERMAL ANALYSIS OF
PARABOLIC TROUGH SOLAR COLLECTORS
FOR TECHNICALLY LESS DEVELOPED
COUNTRIES
TECHNICAL REPORT 1
Prepared by:
Halil M. Given, Research Associate Richard B. Bannerot, Principal Investigator
Approved by:
Prepared as part of the project,
"Medium Temperature, High Efficiency
Tracking and Nontracking Solar Energy
Collectors for Rural and Industrial
Applications" for the Agency for
International Development under the
Agreement for Projects under USAID
Grant Agreement Dated August 26,1978
between the President of India and
USAID for Technologies for the Rural
Poor.
USAID Project Grant No. 3860465
T
une, 1984
Acknowledgements
This work was supported by United States Agency for
International Development (USAID). The assistance given by
the staff of Sandia National Laboratories  Albuquerque, N.M.,
is acknowledged. In particular, cooperation of Jim Banas,
Arthur C. Ratzel and George W. Treadwell in obtaining EDEP
computer code is appreciated.
Halil M. Guven
Richard B. Bannerot
iii
In the U. on the other hand.. rely on imported oil. The use of parabolic trough solar collectors (PTCs. a need exists for low cost alternate energy systems to meet the energy needs of the industrialized as well as the deve loping nations of the world. which are capable of supplying thermal energy over a wide range of iv . and other industrialized countries where solar knowhow and technology exists. Since labor costs in the developing countries are much lower than in the industrially developed countries. abundant in many parts Energy from the sun is of the world and solar energy systens definetely appear to have potential for providing low cost energy. domestic hot water).Executive Summary With the continued escalation of costs for traditional sources of energy. at present solar energy is costeffective for only a limited number of applications (e. the installed cost of a solar system can be considerably less. appears The major to be more favorable in the developing countries. The situation. especially for low to medium temperature applications such as home heating and industrial process heat.S.g. This need is especially pronounced in countries that have no indigeneous oil resources and. As much as 60 percent of the large initial expense of the solar energy conversion systems is for labor. however. share of the cost of solar energy lies in repayment of the initial capital investment for the installed cost of the conver sion system that can transform insolation into a form of energy that can perform useful tasks. consequently.
different designs should be developed for different environments. New analytical models are developed that incorporate the effect of gross errors that may result from poor manufacture/assembly or from labororiented production/assembly or lesser technological capabilities.S. the existing analysis techniques for modeling PTC performance (developed in the U.S. this is a valid assumption for the U. In this report. gross errors in manufacture. It is found that in where most of the PTC research has been performed and the present PTC stateoftheart has been developed. it is concluded 'optimum' PTC that in order to be costeffective. the first of two covering the analysis and design of PTC's.S.temperatures and presently are the leading solar technology in the intermediate temperature range) in developing country energy applications is investigated in this study. the main design 9bjectives have been the encies' and 'maximization of thermal effici These criteria. which favor laborintensive designs and production techniques. assembly and operation were assumed nonexistent in optical models. were found to be incompatible with the goals and design objectives of most developing countries. nonetheless. the U. 'mass production suitability'. Consequently. analytical It is found that existing techniques incorporated restrictive assumptions. mostly) are reviewed and their utilizability in analyzing the performance of PTC designs for developing countries are discussed. For example. design environment as it is a high technology design/production environment.. It is found that these new models can be ly effective used in the design analysis of PTC's for developing country v . however.
The analytical techniques developed herein will be used for developing optimum PTC designs for any given developing country energy application. vi . The second technical report will emphasize the design synthesis of PTC's for developing countries.energy applications.
In such environments. acceptable to choose realistic values for error parameters error tolerances) to carry out the preliminary design of troughs in an efficient manner. The need for detailed knowledge of the effect of various errors and an effective method for determination of realistic error tolerances becomes much more apparent when one looks at the problem of designing parabolic troughs for production and use in technically less advanced design environments such as developing countries. However.e. use of unskilled labor in production may vii . the preferred labororiented production techniques may limit the quality and precision of the final product and result in parabolic troughs with efficiencies. 'lessthanmaximum' optical and thermal Moreover. Employing designs that require no sophisticated maintenance. parabolic trough design goals and objectives may include: Utilizing designs which require small capital investment in new production facilities and depend heavily on local labor.Abstract Proper identification of all possible optical errors (and imperfections) and knowledge of their effects on the optical and overall performance of the trough is vital for the successful design and dimensioning of the trough. enable a designer to determine Such information would 'how much error is too much?' and (i. Utilizing locally available materials and technology with minimum dependence on imported materials and components..
standard deviation of the effective error distribution. intercept aperture that factor. (dr)y is shown that the fraction of rays incident on the collector is intercepted by the absorber (i. potential optical errors are divided into two groups as random and nonrandom errors. mirror nonspecularity. i. absorber diameter)... Three error parameters universal to all collector geometries. which combine random and nonrandom errors with collector geometry parameters are introduced. when designing collectors for developing country environments. ana design and dimension the trough accordingly. where D is A detailed raytrace computer routine which maps rays from elemental reflector surfaces to absorber surface is viii . misalignment of the receiver with the effective focus of the reflector. In the optical model presented here. Therefore.give rise to gross errors that may not be expected in technically advanced design environments. and nonrandom receiver mislocation error. It nonrandom 6. universal error parameters. random error parameters. 3 (= C) and d ) (= (d ry /D. a* (= oC). apparent changes in sun's width.e. The universal error parameters are made up of one and two universal non universal random error parameter. y) is then a function of both random and nonrandom errors as well as the collector geometry parameters such as concentration ratio (C) and rim angle (4). it is increasingly important for the designer to be fully aware of all potential errors and their effects on the performance.e. Small scale slope errors. Reflector profile errors. with builtin tolerances. and small occasional tracking errors are classified as random errors and they are combined into a single random error parameter. a. and misalignment of the trough with the sun are classified as nonrandom error parameters: tracking error.
The optical performance of the trough is The found to be sensitive to both random and nonrandom errors. A comprehensive thermal analysis model is also presented. It is shown that the universal error parameters allow for comprehensive optical analysis of the parabolic troughs.used to validate the existence of the universal error parameters. It is concluded that the results of this optical analysis can be used in comprehensive preliminary design studies for any given design environment. used in a comprehensive design and optimization method. the effect of the random and nonrandom errors on the performance of the trough can be summarized in universal design curves in the form of intercept factor versus universal error parameters for a given rim angle. It is shown that this model can be used to calculate a heatloss parameter qL in Watts per m 2 of receiver surface area to characterize the thermal behavior of the receiver. A onedimensional heat transfer model for the thermal analysis of the receiver subsystem is adopted. The optimum rim angle based on both random and nonrandom errors is found to be in the range of 1050 to 1200. optimum rim angle based on random errors alone is found to be broad (in the range of 800 to 120'). It is shown that the presented thermal analysis can be The method developed can be used to size the annulus gap size. ix .
1 3.2 Industrialized versus Developing Countries 1.4 Description of Optical Errors 3.OPTICAL AND THERMAL ANALYSIS OF PARABOLIC TROUGH SOLAR COLLECTORS FOR TECHNICALLY LESS DEVELOPED COUNTRIES TABLE F CONTENTS O Page Acknowledgements Executive Summary Abstract Table of Contents iii iv vii x List of Symbols List of Figures List of Tables Chapter 1 GENERAL INTRODUCTION xii xvi xxi 1 1 1.2 3. ASSEMBLY AND OPERATIONIntroduction Optical Efficiency Modeling Intercept Factor for OnAxis 4 5 8 11 Chapter 3 15 15 16 3.1 Background 1.3 Operation 3.5 Analytical Techniques for Modeling Incident Sunshapes and Effective Sunshapes 17 21 29 x .3 Literature Survey and Dicussion 1.4 Objectives and Outline of the Study Chapter 2 OUTLINE OF A COMPREHENSIVE PERFORMANCE SIMULATION MODEL FOR PARAEOLIC TROUGHS METHODS FOR MODELING ERRORS IN MANUFACTURE.
104 106 106 107 110 116 118 121 122 124 6.1 Solar Irradiance Profile  Incident Sunshape 3.4 Calculation of HeatLoss Parameter qL 6.5.2 Reflected Beam Profile  Effective Sunshape 3.1 PRELIMINARY OPTICAL RESULTS AND DERIVATION OF UNIVERSAL ERROR PARAMETERS RESULTS OF THE COMPREHENSIVE OPTICAL ANALYSIS Introduction 53 78 78 5.5.3 OneDimensional HeatLoss Model 6.3 Results of Optical Analysis with Random and Nonrandom Errors 5.Optimal Rim Angles 5.5 Chapter 6 6.Page 3.1 Summary THERMAL ANALYSIS OF PARABOLIC TROUGHS Introduction 79 82 98.7 Summary 29 31 38 39 42 50 52 Chapter 4 Chapter 5 5.5.2 Results of Optical Analysis with Random Errors Alone 5.6 Effect of OffAxis Operation 3.2 Description of the Problem and Assumptions 6.4 Sensitivity of Error Tolerances to Changes in Rim Angle .5 6.6 Chapter 7 REFERENCES xi Annulus Gap Sizing Summary CLOSURE .3 Analysis of Optical Errors Statistical Error Analysis Comprehensive Error Analysis 3.
a numerical value asso lated with an attribute of a particular alternative Amax i n Maximum attribute rating of attribute j Amin Ar C C0 CO <cose> Cp D D gi Minimum attribute rating of attribute j Absorber tube area Concentration ratio.LIST OF SYMBOLS Aa Reflector aperture area attribute rating for the i h alter naAive. C = W/iD Optimum concentration ratio Optimum concentration ratio for sensitivity analysis Allday average incidence factor Specific heat at constant pressure Absorber tube diameter Diameter of inner surface of glass jacket (glazing) Diameter of outer surface of glass jacket (glazing) Distance between the actual focus and the center of the absorber tube D )0 dr (dr) rx d r along the lateral axis of the reflector (d )y d* d+ i d r along the optical axis of the reflector Universal nonrandom error parameter due to receiver mislocation (= (d )y/D) Amount by which constraint i has exceeded goal (overachivement) xii .
3. their subscripts serve to identify priority level for the goals Heat loss per receiver surface area qL xiii .List of Symbols (Continued) Amount by which constraints i is short of goal (underachivement) Focal length of the reflector Acceleration due to gravity Heattransfer coefficient due to wind Ratio of diffuse to hemispherical Beam solar irradiation Allday average beam insolation Allday average beam insolation and incidence factor product Thermal conductivity of air Effective conduction coefficient due to conduction and natural convection in an annular space kgas Thermal conductivity of the gas in the annu lar space of the receiver Thermal conductivity of glass Sky clearness index Incidence angle modifier (see Eq. f g hw Hd/Hh Ib <Ib> <I b cosa> k air kef kglass Kh K(e) <K(e)> zAnnular L 1n Pi gap width Absorber tube length Working fluid flow rate Priorities of goals.2) Allday average incidence angle modifier irradiation d.
o UL Vwind W Greek Symbols a Absorptance of selective coating Reflector misalignment and tracking error angle (see Figure 3.i Tg.3) Universal nonrandom error parameter due to angular errors (= aC) y to Intercept factor at normal incidence Optimum intercept factor Instantaneous (incidence angle dependent) intercept factor 5Solar declination angle xiv .List of Symbols (Continued) qnet <qnet> Net heat flux absorbed by the absorber Allday average net heat flux absorbed by the receiver Q R ij Heat loss rate Normalized jth attribute rating for the ith alternative Ambient temperature Absorber temperature (collector operating temperature) Inlet temperature of working fluid Outlet temperature of working fluid Temperature of inner surface of glazing Temperature of outer surface of glazing Heatloss coefficient for receiver based on absorber surface area Wind velocity Reflector aperture width Ta Tabs Tfin Tfout Tg.
o * (Ta) n 4Trough w xv .3.List of Symbols (Continued) Eabs Eg.1) (see 6 61.3.2. 6c Incidence angle projected to yz plane (see Fig.1) Incidence angle on the aperture at the cutoff time enoon xLatitude 11 p Incidence angle at solar noon Angular shift of the mean of the standard distribution (see Eq.o To <rlc> (TC)noon 6 Optical efficiency (see Eq.3.3.i Emissivity of selective coating Emissivity of glazing inner surface Emissivity of glazing outer surface Half optical cone angle (angular aperture) Eg.18) Average specular reflectance of the reflec ting surface Energy distribution standard deviation Universal random error parameter (= aC) Transmissivity of the glass jacket Effective transmittanceabsorptance factor at normal incidence rim angle (see Fig.1) Hour angle (subscript c designates hour angle at cutoff time).1) Allday average collector efficiency Overall collector efficiency at solar noon Angle of incidence of the sun's rays on the collector aperture measured from the normal to trough aperture Incidence angle projected to xy plane Fig.
.................1 3..............6 Schematic Representation of a Mirror Surface Showing the Difference Between Slope Errors and the Mirror Diffusivity ................................2 Idealized Central Ray Optics of Parabolic Troughs ............................ 27 3............... 2 2....3 3......................1 Common Terminology of Parabolic Trough Collectors.. 18 3.............. 12 Definition of Coordinates and Projected Incidence Angles for TwoDimensional Reflectors .. 22 Schematic Representation of a Parabolic Mirror Surface Showing the Difference Between Slope Errors and Reflector Profile Errors ........................................7 Abstract Representation of Scattering by Reflection ... 32 xvi ...........1 Title Temperature Ranges Attainable with Different Page Solar Technologies........ 20 Description of Potential Optical Errors in Parabolic Trough Collectors ....24 3........................... 25 3.4 3...... ............LIST OF FIGURES ligure 1.......................................5 Ray Traces of Reflection from Perfect and Imperfect (Random) Mirror Surfaces Showing Perfectly Reflected Central Ray sn and Imperfectly Reflected Central Ray s..
....... 44 3. ...... = 4.... a Intercept Factor as Function of Receiver a Mislocation Along the Lateral Axis and Tracking Error for = 0..8 (Continued) Title Page Modeling of Potential Optical Errors in Parabolic Trough Collectors ......01 radian ...... ..4 Intercept Factor as a Function of Receiver Intercept Factor as a Function of Receiver 59 4.. ...... 67 4....... a islocation Along the Optical Axis and Tracking Error for a 0.... 4...9 Intercept Factor as Function of Concentration a Ratio and Random Error Parameter ( Zero Non Random Errors) ............... a Intercept Factor as a Function of Receiver Mislocation Along the Lateral Axis and Tracking Error for = 0.......... 60 4......10 xvii ..1 55 4...List of Figures Figure 3..................... ..02 radian ...508 cm ..0025 radian .. Intercept Factor as a Function of Receiver 61 4.9 Combination of Receiver Location and Parabola Profile Errors .016 and 1................3 57 4...................... ...01 = radian ...5 Mislocation Along the Optical Axis and Tracking Error for a 0.....02 radian . 49 Intercept Factor as a Function of Receiver Mislocation Along the Lateral Axis and Tracking Error for = 0... ....424 cm .6 Mislocation Along the Optical Axis and Tracking Error for a = 0..8 Intercept Factor as a Function of Tracking Error and Random Errors for (d ) = 1...0 and and 0...........2 56 4.................... Intercept Factor as a Function of Tracking 63 64 4..... 65 Intercept Factor as Function of the universal a Random Error Parameter (Zero NonRandom Errors)..................0025 radian .....7 Error and Random Errors for (d ) = 0.
......27 cm ........18 Intercept Factor as a Function of the Universal NonRandom Error Parameter due to Receiver Mislocation ...... for * = 0. .........0 ...... Allowable NonRandom Error Levels for y = 0.......54 cm ..05 radian and (d ) = 0......................4 84 Xviii ........0 ...... 5...16 Intercept Factor as a Function of Concen tration Ratio and Receiver Mislocation for D = 2....2 Intercept Factor Versus Rim Angle and the Universal Random Error Parameter (Zero Non Random Errors) .10 radian and (d ).......3 5....... = 0.0 .............. 4.......1 Intercept Factor as a Function of the Universal Random Error Parameter and rim angle (Zero NonRandom Errors) .................. 4..................17 Intercept Factor as a Function of Receiver Diameter and Receiver Location Error ...15 Intercept Factor as a Function of Concen Ratio and Receiver Mislocation for tration 72 D = 1....13 Intercept Factor as a Function of Concen tration Ratio and Tracking Error for a* = 0...... ry 4... 5.. 4..14 Intercept Factor as a Function of the Universal NonRandom Error Parameter due to Angular Errors (8*) and Various Random Error Levels (Zero Receiver Location Error) .20 radian and (d ) = 0..................12 Intercept Factor as a Function of Concen tration Ratio and Tracking) Error .. 68 .........................11 Intercept Factor as a Function of Concen tration Ratio and Tracking Error for a* = 0... 73 4.95 and C = 800 and Various Random Error Levels .... 31 Qualitative Representation of Intercept Factor as a Function of the NonRandom Error Parameters for Fixed Rim Angle and Various Value of Random 83 Error Parameters .......... 74 75 80 5............... 70 71 4......... 69 4..List of Figures Figure (Continued) Title Page 4.......
. 100 5..75 and 0 = 1200 and Various Random Error Levels ..18 Allowable Nonrandom Error Level d as a Function of Rim angle and i* for a* = 0...2 rad ..........1 rad ......95 and = 90' and Various Random Eror Levels . 5..95 and 0.95 and = 1200 and Various Random Error Levels .................9 89 5.... Allowable NonRandom Error Levels for y = 0....13 Allowable NonRandom Error Levels for y = 0..10 Allowable NonRandom Error Levels for y = 0.....35 rad.........75 .. 5...7 87 5.75 .15 Effect of Rim Angle on the Allowable NonRandom Error Levels for a* = 0...'75 ..75 ...... 5.12 Allowable NonRandom Error Levels for y = 0.6 86 5...75 and 0 = 100' and Various Random Error Levels .. Allowable NonRandom Error Levels for y = 0. and y = 0...75 and T = 1100 and Various Random Error Levels .. 101 xix ....... 5..5 (Continued) Title Page Allowable NonRandom Error Levels for y = 0.75 and 0 = 90' and Various Random Error Levels ...11 Allowable NonRandom Error Levels for y = 0. 5..14 Effect of Rim Angle on the Allowable NonRandom Error Levels for a* = 0.2 and for y = 0.... 90 91 92 93 95 96 97 5.... 99 5......... Allowable NonRandom Error Levels for y = 0........................ Allowable NonRandom Error Levels for y = 0.19 Allowable NonRandom Error Level d* as a Function of Rim Angle and B* for a* = 0.17 Allowable NonRandom Error Level d as a Function of Rim Angle and * for a* = 0.....................75 and 0 = 800 and Various Random Error Levels .List of Figures Figure 5. 85 5....8 88 5.. .. 5...95 and = 1000 and Various Random Error Levels ..4 and y = 0.16 Effect of Rim Angle on the Allowable NonRandom Error Levels for a* = 0..95 and 0 = 1100 and Various Random Error Levels ...3 and y = 0.... ...................
............... Wind Induced HeatLoss from Receiver ... 5.................. HeatLoss Coefficient U as a Function of Average Absorber Tube Tmperature and Absorber Tube Diameter ....75............. Page 5.7 xx .......... il 115 117 6...........4 6....................... TwoDimensional Energy Exchange to the Working Fluid [40] ............List of Figures (Continued) Figure Title ...............2 6............3 6.20 Allowable Random Error Level d as a Function of Rim Angle and 6" for a*= 0.0 and y = 0..95 and 0..........75 ...... 6.0 and y = 0...... Annulus Gap Sizing for an Absorber Temperature of 200 0C ..54 cm Absorber Tube for an Absorber Temperature of 315'C... 102 103 108 108 OneDimensional Receiver HeatLoss Model ......5 Cross Section of Receiver Subsystem . Annulus Gap Sizing for Alternate Gases using a 2..95 and 0..............21 Allowable NonRandom Error Level 8 as a Function of Rim Angle and a* for d*= 0..1 6....6 120 121 6..
1 3.2 28 45 6.LIST OF TABLES Table Title Summary of Potential Errors Typical Focal Lengths and Slope Errors for Model Parabolas of Various Materials and Processes Reference Trough Receiver Design Parameters Page 3.1 117 xxi .
the U. chemical fertilizers and hybrid seeds. and to provide for industrial development through rural electrification and road construction. lulled the world into a false sense of energy security. interest in solar energy and other alternative energy sources was intensified and the development of indigenous fossil fuel resources.2] Figure 1. conservation programs. and other industrialized nations of the west based their entire economic growth on the nonrenewable energy resources and the developing countries of the world followed their lead by initiating many fossilfuelintensive development programs to accelerate agricultural production through the use of irrigation. inipcovements in energy conversion and efforts to utilize renewable sources were initiated.1 Background The continuing expansion of fossil fuel reserve estimates in the 1960's. feasibility of many alternative energy op The attainable temperature ranges for various technically . where solar knowhow and !umbers in brackets refer to references in Reference section.S. This intensified research and development in renewable energy sources resulted in the demonstra tion of the technical tion.S. * are shown in Nonetheless.c. Nevertheless. the energy crisis of 1973 and the subsequent tenfold increase in oil prices has greatly decreased the attractiveness of many fossilfuelintensive development and economic growth programs in both industrialized and developing countries. . feasible solar energy utilization systems [1. In response to the energy crisis./ Chapter One GENERAL INTRODUCTION 1. During this period.1. in the U.
Type of collector: 4000 F 2000O F
Central receiver
Point focus (parabolic dish and fresnel lens) 1504F Line focus (parabolic trough and fresnel lens, also
multiple reflictor)
4000 F
20000 F
600OF
120F Evacuated tube 160'F Flat plate
180OF
3509F
Solar pond
50 I 10
100 I 50
200 I 100
400
800
1000 I 500 I
2000
3000 I 2000
(OF) ('C)
1000
Operating temperature
Note:
Linefocus, evacuated tube, and flatplate collectors are commercially available; central receivers, polntfocus collectors and solar ponds are still being developed.
Figure 1.1
Temperature ranges attainable with different solar technologies (from Reference 2).
3
technology exists, at present, solar energy is costeffective
for only a limited number of applications (e.g., water) system, [2,3]. is domestic hot
The economic value of a solar energy utilization to the incremental cost of the cheapest
equivalent
competing energy source that supplies the energy to accomplish the same task as the solar system. At the present time with
government orchestrated energy p,'icing policies, taxex emtptions, etc., especially favorable to conventional energy,
it is difficult to make a true cost estimate. However, rough
estimates of collection efficiencies, material costs, heat
exchange efficiencies, etc., recently indicate that in the
U.S., for example, thermal energy from a solar system would
cost more than that from a conventional source.
The situation, however, appears to be more iavorable in de veloping countries. The major share of the cost of solar energy
lies in repayment of the initial capital investment for the installed cost of the conversion system that can transform insolation
into a form of energy that can perform useful tasks. as 60 percent of the large initial As much
expense of the solar energy Since labor costs in the
conversion systems is for labor [4].
developing countries are much lower than in industrially devel oped countries, siderably less. the installed cost of a solar system could be con Furthermore, since many developing countries have the price of the cheapest competing
no indigenous oil resources, energy source,
which is usually imported oil, is considerably the U.S. Thus, a solar industry,
higher than, for example, in
4
utilizing local labor and materials as much as possible, may
be costeffective today or in the near future for many appli cations in developing countries.
1.2
Industrialized versus Developing Countries
Of the existing solar technologies (Figure 1.1), parabolic
trough and flat plate collectors have a moreorless fully devel oped technology, and they are commercially available in the U.S.
Parabolic Trough Collectors (PTC's) are capable of supplying
thermal energy over wide range of temperatures (up to about
a 300*C),and therefore they can be used for a variety of applica tions ranging from electricity for rural applications (agricul
tural pumping) all the way to industrial hot water and steam
production. Consequently, at present, they are the best can
didates for widespread use in developing country energy appli cations.
In mounting an effort to encourage the utilization of PTC's
in developing country energy applications, however, it is
important to be fully aware of the differences in design
philosophy and design objectives between industrialized and
developing countries. the U.S., In an industrialized environment like
design goals and objectives for PTC's included [5]:
High performance  thermal efficiencies of 6070%
at 600OF (305 0 C);
,
Throughout this text 'environment' is defined as a country (or
region) with certain given economic and technical capabilities.
N. 'information base' However. • Lowcosts achieved with lowlabor..g. On the other hand. locally available materials and technology with minimum dependence on imported materials and components. massproduction materials and processes. The problem is then to satisfy the goals and objectives of designs in developing countries with an derived from industrialized countries. Albuquerque. an important question that needs to be answered first is: Is the current 'information base' in PTC technology comprehensive and flexible enough to be used in developing PTC designs su'itable for developing country energy applications? The answer to this question is given in the next section.5 • Designs suitable for massproduction. designs that require no sophisticated maintenance.. Sandia National Laboratories.3 Literature Survey and Discussion An extensive search of the literature and communication with the researchers in the world's leading laboratory in PTC technology.M. 1. . in a semiindustrialized and technically less advanced environment (e. design goals and objectives for PTC's include the use of: designs which require small capital investment in new production facilities and raw material and which depend heavily on local labor. developing countries).
research efforts were invested in the development of one for all applications and geographic U. 'standardized' PTC design locations throughout the 'standard' PTC design concept is For example. in a study investigating the influence of geographic location on the PTC performance. emphasized Treadwell in almost all the available studies.6 unveiled a wealth of scientific information on PTC analyses and performance evaluations (e. [614]). The 'transportable'. These included numerous studies on optical and thermal analyses as well as studies on manufacturing tehniques. it was concluded that stateoftheart PTC modeling was not comprehensive enough to be utilized for performance simulations in a comprehensive design method which would incorporate changes ii. etc.S. [14]. a singleaxis tracking parabolic trough solar collector could have a common optimum design for use in all regions of the U. After carefully studying the available literature.S. As a direct consequence of the ' lowlabor massproduction goal of the U.. wind loading. the design environment. flexible hose design. for the deficiencies that existed in the PTC analysis models at the beginning of this investigation are discussed in the following paragraphs. the answer to The reasons the question posed in the preceding section is no. Therefore.S. design environment. This determination has an impact upon mass production of troughs since different .g. materials improvement. stated: ". selective coatings... methods for reducing slope errors.
Instead. and could result in multiple production lines and controlled distribution. this design objective led to the development of performance simulation models which were almost specifically tailored for design analysis of a very limited number of PTC geometries. different eometric configurations locations (just the can be considered for different opposite of the ONE optimum design philosophy ." Subsequently.. Therefore. In developing country design environments: Big massproduction facilities and largevolume production (which require efficient transportation distribution systems too) may be out of question. Neither is necessarily compatible with the desired cost reductions that could result from large volume production. Sandia researchers chose a 2maperture width 90°rim angle reflector and a 2.g. For example. A fundamental study to develop comprehensive mathematical models which could be used for all collector geometries was not undertaken.14]).7 geometric configurations might be optimum for different regions of the U.S. 'distributed production shops' may play a more important role in costeffectiveness and hence widespread use of PTC's. smallscale laborintensive. [13.5cmdiameter cylindrical receiver with selective coating for their quality' PTC design in 1976 [15] and this design 'high formed the basis for all of their subsequent studies and modeling (e.
Therefore. The preferred labororiented production techniques and lesser technological capabilities will limit the quality and precision of the final product and also introduce gross optical errors. (In the gross errors in manufacture. assembly and operation were assumed nonexistent.8 stated above). the effect of such errors should be accounted for in the design calculations. Nonetheless. when designing PTC's for developing country applications. this is a valid assumption for the U. . To develop a comprehensive PTC optical model which incorporates the effect of gross errors.) 1. stateoftheart PTC optical model [18]. design environment as it is a high technology design/production environment. Therefore.4 Objectives and Outline of the Study The main objectives of the work to be presented in this report can therefore be stated as follows: 1.S. in such environments 'redesigning' PTC's for different anplications and locations may be a must to compensate for perfor mance reductions resulting from poor manufacture and/or poor maintenance. In fact. it is desirable to have simulation models that can analyze different collector geometries.
typical results from the comprehensive optical model developed in Chapters Three and Four are presented. B* and d*) are introduced to charaterize the various potential errors. errors are classified as random and nonrandom errors.9 2. . deficiencies that existed in the PTC The specific 'information base' at the beginning of this study are discussed. meters (one random. In Chapter Two. which can be used to analyze different collQ'or geometries and which can eventually be used in comprehensive design studies to develop 'optimum' PTC designs for developing country design environments. error parameters universal to all collector Potential geometries are derived and preliminary results from the optical model are presented. the nature of the gross optical errors in manufac ture. The universal error parameters enable the designer to analyze different geometries. the outline of a 'comprehensive' perfor mance simulation model for PTC's is given. That is. including both optical and thermal effects. In Chapter Three. the first step in the development of a comprehensive optical model for the PTC's is presented. In Chapter Five. and two nonrandom. The new error parameters are called Three universal error para 'universal error parameters'. To develop a comprehensive performance model. In Chapter Four. a*. assembly and operation are described and techniques are developed for modeling the gross (systematic) errors.
Concluding remarks. It is concluded that the optical model developed is comprehensive as well as versatile and therefore it can be used in a comprehensive design method to develop optimum collector designs for developing countries.10 The model is validated by comparing the results with those of the previous investigators. Available literature on thermal analysis of PTC's is reviewed and the use of these studies in a comprehensive PTC design method is discussed. In Chapter Six. discussions and recommendations are given in Chapter Seven. . analysis of the thermal behavior of PTC's is presented.
The receiver an absorber surface and The collector tracks the sun on a continuous basis focusing the sun's rays onto the absorber surface which becomes heated and transfers energy to the fluid flowing through it. decoupled and hence dealt with separately properties of the materials used are assumed to be temperature independent. undoubtedly. 11 . can be carried out completely independent of each other and this. The heat transferred to the fluid can provide energy for many types of practical energy applications. this is an excellent approximation. logy for PTC's. a receiver assembly centered along the reflector's focal line and a tracking mechanism. A PTC performance simulation model consists of two parts: an optical analysis for the reflector system and a These can be if the optical Figure 2. Therefore. optical and thermal. including PTC's. assembly usually has two components: a transparent outer cover (glazing).I/ Chapter Two OUTLINE OF A COMPREHENSIVE PERFORMANCE SIMULATION MODEL FOR PARABOLIC TROUGHS A PTC consists of a cylindrical parabolic reflecting surface (reflector). It turns out that in most solar thermal systems. brings about a considerable simplification in the performance simulation modeling.1 presents the common termino thermal analysis for the receiver system. the two analyses.
12 Sun's ray Cylindrical Receive ' \ I. .< ! '"Glazing Reflector wD W W' f = Focal length = Rim angle W = Aperture width D = Receiver diameter 0l = Acceptance angle = Gap width Figure 2.1 Common terminology of Parabolic Trough Collectors.
which are more of a rule than exception in low technology environments.g. The stateoftheart optical models for PTC's were found to be incomplete (i.e. however. were assumed to be nonexistent.. however.ing (e. In the stateoftheart optical mod. As discussed earlier. hightechnology design environment where a microcomputer controlled tracker. one cannot find a comprehensive 'information base'. Gross (largescale) systematic optical errors.too.. . have restrictive assumptions) and hence were not directly adaptable to a comprehensive PTC design method.. With regard to the optical analysis of PTC's. the major deficiency was found to be in the handling of the optical errors. When designing PTC's for lowtechnology design environ ments.. the optical errors were treated as random processes and their occurances were represented as normal distributions with zero mean.g. [16]). the effect of gross errors in manufacture. precise manufacturing techniques and sophisticated quality control equipment (e. one can find a rather well developed literature which includes extensive experimental data. trace [17] laser ray for assessing the optical quality of the reflector) can be used.13 With respect to the thermal analysis of cylindrical receivers (with no insulation) . These studies are reviewed and their utilizability in a comprehensive PTC design method are discussed in Chapter Six. This assumption seems to be a valid one for an industrialized.
* determine the minimum concentration ratio (relative sizes of reflector aperture and absorber tube. * determine tradeoffs between accuracy and quality of reflector surface (e. and between concentration ratio and absorber coating properties. Therefore. model the effect of gross (systematic) optical errors. it was concluded that studies were needed to: .. slope errors and reflectivity) and efficiency.14 assembly and operation should be incorporated into the collector simulations and accounted for in the design calculations. The following chapters present the development and results of comprehensive optical and thermal analysis models . W/frD) that is capable of delivering energy at a specified temperature for given optical error levels.g.
imperfect tracking of the sun. imperfections (or errors) that may result from poor manufacture and/or assembly. In this chapter procedures are presented for modeling the effect of various manufacturing. and poor operating procedures. * physical properties of the materials used in receiver and reflector construction.and hence methods are developed for calculating the intercept factor for different trough geomet ries having different error levels. Proper identification ol all the factors that affect the optical performance and hence precise knowledge of their effects are vital for the successful design and dimensioning of the trough. the average concentration ratio of a trough is degra ded to values much below this upper limit due to: * apparent changes in sun's width and incidence angle effects. ASSEMBLY AND OPERATION 3.1 Introduction The upper limit to the concentration that a parabolic trough can achieve is set by the sun's width. In practice. however. .Chapter Three METHODS FOR MODELING ERRORS IN MANUFACTURE. 15 . operational and material imperfections (errors).
1) Average specular reflectance of the reflective surface. Tj The optical efficiency. the instantaneous intercept factor Ye decreases with incidence angle. These factors include the inci dent angular dependence of glazing transmittance and absorp tance. the incidence angle effects can be separated out . given by Eq. This decrease is brought about in two First. e= Angle of incidence of the sun's rays on the collector aperture.. ways. angle of . the apparent path length.2 Optical Efficiency Instantaneous optical performance of a trough can be measured by the optical efficiency.16 3. there is beam spreading due to reflector longi errors.1). Second.(3. Ye = Instantaneous intercept factor (defined as the fraction of rays incident upon the aperture that reach the receiver for a given incidence angle e). varies with incidence between the aperture surface normal and the There are several factors incoming radiation. Also. which is the fraction of incident radiation absorbed by the absorber tube and is expressed as: no = P(Ta) ef f Y where P = (3.7. (Ta) ef f = Effective transmittanceabsorptance factor for the receive. that contribute to the decrease of optical efficiency with increasing incidence angle. tudinal slope and nonspecularity sun image becomes wider due to the longer reflected Fortunately.
modifier K(e) defines how the optical efficiency decrease with incidence angle. operation and materials imperfections/errors.(3.1.1) can be rewritten as: no = [K(e)][P(Tcz)n] Y (3.4) This definition of the optical efficiency allows a clear distinction of the factors contributing to it. Therefore.3 Modeling Intercept Factor for OnAxis Operation In Figure 3. contains the effects of various manufacturing. The second bracketed term represents the optical effects of the physical properties of the materials used in receiver and reflector construction. The first bracketed term. The incidence angle modifier at a given angle K(O) E e is defined as [18]: 0( .2) where. y . 3. Eq. the coordinate system chosen for the .0)e [P(ra) effy6](8) (3. Y = Intercept factor at normal incidence.3) where (Ta)n = Effective transmittanceabsorptance factor at normal incidence. accounts for all the incidence angle related effects. relative to the trough's normal incidence optical efficiency. The last term. the intercept factor. (no)n is the normal incidence optical efficiency and it is expressed as follows: (qo)n = [P(Ta)n] Y (3.17 by using the socalled incidence angle modifier [18] . The incidence angle K(e).
." I / (b) Y x"   IY Figure 3.18 (a) • ".xand (b) e 1 1 for twodimensional reflectors (from Ref.1 Definition of coordinates and projected incidence angles (a) S. 16).
there are no errors in the construction and operation of the system nor optical losses. In this ideal case the central ray from the sun is incident along the direction of the arrow in the Figure 3. y is defined as a purely geometric quantity with out regard to absorption. the reflectance transmittanceabsorptance product tion (3. The z axis is placed along the tracking axis. Intercept factor is defined as the fraction of the rays incident on the aperture that are intercepted by the receiver. A light ray would reflect from the trough at P and go toward the focal point 0 of the parabola.1). absorption losses will be accounted for later by a multiplicative factor. as shown in Equa could be considered where the sun is a point source. an "ideal" (but nonexistent) situation ('TC ). In this work. In this section models are presented for the calculation of intercept factors for onaxis operation. and y axis along the axis of symmetry or optical axis. 3.19 optical analysis of the trough is shown. For simplicity.2.. respectively. The collector is said to be onaxis when the incidence angle is zero. i. 0 and ell are the projec tions of the incidence angle of the sun on the xy plane and on the yz plane. In this case. the intensity distribution of both incident and reflected rays would be . With perfect tracking the misalignment angle 11 vanishes and el1 equals the incidence angle. the direction of the incident central ray from the sun is perpendicular to the aperture plane of the trough.e. there is no atmospheric (aureole) effects. as shown in Fig.1.
20 Y Incident central Circular absorber x rayf Reflector Figure 3. .2 Idealized central ray optics of parabolic troughs.
a statistical description is developed which will enable one to calculate the probability that the ray will reflect into a specified interval of directions which is adequate for calculating averages. Finally. The sun's rays do not arrive at the site as a collimated beam but as a somewhat dispersed beam.3 presents a schematic representation of various types of potential errors that may be encountered . and it would be a simple matter to analyti cally calculate the fluxdensity pattern on the absorber tube and to determine the intercept factor.4 Description of Optical Errors Figure 3. In the real world. numerical evaluation of the intercept factor using the statistical descriptions is presented. Moreover. detailed descriptions of potential optical errors in the trough construction and operation are presented. life is not so simple. 3.21 delta functions. Therefore. statistical characterization of the sunshape and errors are presented and discussed. however. even the central ray coming from the center of the solar disc may not reflect in the ideal manner but deviate from it because of slope errors on the trough near P or because of suntracking errors. first. Then. The exact direction of the reflected ray is not known because a deterministic description of the direction of the surface normal at P is not available. In the following.
= tracking errors + Effective Focus Nonuniform incident beam intensity Receiver Location Errors \% / "" Material. i Dif fusivity A Parabola Profile Errors y Local slope errors (surface waviness) Figure 3.3 Description of potential optical errors in parabolic trough collectors.22 Concentrator misalignment and '. .
the reflecting composed of a thin sheet When this are considered. Finally. surface of the trough is assumed to be of material. These can be listed as: errors asso ciated with the reflecting surface. First.4. they can be the concentrated image at the receiver. smallscale structure consisting of a grainy texture can be characterized as a material property. and misalignment of the collector aperture with respect to the sun (e.g.23 in parabolic troughs. the actual mean surface deviates from the ideal in such a way as to displace the effective focus.4 .5. (ribs). nonspecu larity (diffusivity) of the reflective material. As shown in Figure 3. A local slope error primarily dependent on the deviation of the actual wavy surface from its mean contributes a deflection which approximates twice the angular deviation between the actual surface from the mean as is shown in Figure 3. mislocation of the receiver with the effective focus of the reflecting surface. one surface of which is reflective. sheet is attached to its supporting structure distortions occur in the surface.. the errors associated with the reflecting surface As a hypothetical example. All these factors will contribute to the blurring of However. namely. scale structure consisting of a grainy texture plus a striation pattern. characterized as three basically independent modes of reflector error. tracking errors). various The resulting surface has a wavy pattern and in general the mean surface obtained by avera ging out the waves also may differ from the ideal (desired) Finally. the reflecting surface may havre a small surface. Figure 3.
24 S Central ray no n 2. .L Actual SMean Ideal / 0' Surf ace 0Ideal Focus Focus of Mean (Effective Focus) Surface Surface Figure 3.4 Schematic representation of a parabolic mirror surface showing the difference between slope errors and reflector profile errors.
25 S nI n SI /  S n P "Imperfect Perfect surf ace N N surface Figure 3.5 Ray traces of reflection from perfect and imperfect (random) mirror surfaces showing perfectly reflected central ray sn and imperfectly reflected central ray s. .
* sagging of the receiver between supports which themselves are elastic and will be distorted by thermal gradients. * thermal growth and expansion of the receiver and its insulation during operation. Table 3.and Figure 3. the projected central ray from the sun may strike the reflector aperture plane at an angle (instead of striking the plane perpendicularly).1 presents a summary of the discussed imperfections/ errors according to their source of origin. Two additional sources of error are those associated with the alignment and positioning of the receiver with respect to the expected focus and with the tracking accuracy of the collec tor drive system.26 illustrates the difference between slope errors and reflector profile errors. This will be due to the rotation of the vertextofocus axis of the reflector during assembly or instantaneous misalignment of the reflector with the sun (tracking errors). As shown in Figure 3.3 at any given time the center of the absorber tube may be positioned at a distance away from the effective focus of the reflector due to: * errors in positioning the receiver tube during assembly. Furthermore.6 illustrated the difference between slope errors and mirror diffusivity. * change in the effective focus of the reflector during operation. at a given time. .
.6 Schematic representation of a mirror surface showing the difference between slope errors and the mirror diffusivity.27 S Central ray n ~Actual surface Ideal surface Figure 3.
3).1 SUMMARY OF POTENTIAL ERRORS MATERIALS . (ii) permanent expansion of the receiver as a result of thermal cycling over a period of time. reflector may be rotated (or twisted) about the vertextofocus axis during assembly (see Figure 3. . Nonspecularity (or diffusivity) of the reflective surface may increase with time. (It may also develop after collector has been in operation over a period of time.3) OPERATION Tracker epuipment may cause tracking bias/error due to its poor quality or tracking biases may develop after the collectors have been in operation for some time. Misalignment of the receiver with the effective focus may develop during operation due to one or a combi nation of the following: (i) sagging or buckling of the receiver tube because of thermal expansion (if insufficient thermal expansion tolerance exists in the design). Profile errors: Average shape of the reflector (ob tained by averaging the local slope errors or waves) may differ from a parabola. temperature effects. may be misaligned with respect to the effective focus of the reflector during manufacture and/or assembly (Figure 3. That is. during operation.. This may be due for example to distortions during manufacture and/or assembly. due to weathering or accumulated dust on reflector. The receiver tube • Mislocation of the receiver tube. Profile errors may develop or increase due to wind loading.28 TABLE 3. (iii) change in location of the effective focus. Nonspecularity (diffusivity) of the reflective material MANUFACTURE and ASSEMBLY • Local slope errors (surface waviness) of the reflector that may result from distortion of its surface dur ing manufacture.) • Misalignment of the reflector during assembly. etc. due for example to increased profile errors in the reflector.
This variation in intensity is directly observed and presumably arises from the interaction of complex mechanisms in the emitting solar atmosphere as well as atmos pheric scattering in the terrestrial atmosphere. the sun later studies [ 8 . In the present work.. i. The direc tional distribution is widened by atmospheric scattering es pecially during hazy atmospheric conditions. the sun is not a pointsource and the solar In recognition of the finite size of assumed a finite uniform inten source.g.5. But. the sun was assumed to be an infinetely remote point However. Light clouds can [20]. disc has a finite size. Scat cause considerable broadening of this distribution tering at a reflecting surface will cause a further modification of the sun's intensity distribution..e. Therefore.1 Solar Irradiance Profile .5 Analytical Techniques for Modeling Incident Sunshapes and Effective Sunshapes 3. a nonuniform distribution of intensity exists across the visible solar disc.Incident Sunshape In the early optical analyses of parabolic troughs (e. 19] sity solar disc. describes the distribution A probability density function )f the sun's rays with respect to the direction of the central ray from the sun and is called the . this is not a realistic assumption A random photon from the sun is drawn from a distri bution of directions and posseses different intensities depending upon where it originates on the sun. in exponential function is used to describe the solar intensity .29 3. either. [ 6] and [ 9]) the sun's rays were assumed to be parallel.
22]. the radiant intensity of the sun's rays for a total incident flux of Ib is described by a normal (Gaussian) distribution function as follows [21]: Isun(O = 2 7T exp[sun 2 2 02 ] (3. .5) sun where asun is the scattering parameter (standard deviation of the scattered distribution) and . Therefore.6 mrad. The optical cone angles () may be either one or two dimensional for parabolic trough or heliostat methods of energy collection respectively. is the half optical cone angle (angular aperture). This representation of the sun's rays intensity distribu tion is particularly adaptable to treating the random scat tering at material surfaces or effective scattering induced by manufacturing errors and tolerances. of these distributions ranges from a minimum value of 3.30 "sunshape" [21.55mrad to a maximum of 22. A set of sixteen "standard" sunshapes generated from measurements performed by Lawrence Berkeley Laboratory [23] are available in the litera ture. Approximation of the intensity distribution of the sun across the solar disc by a normal probability distribution is found to be adequately accurate for optical analysis of troughs when optical errors are large compared to the width of the sun [16]. These sunshapes are intended to represent a wide range The RMS width of shapes of interest to solar applications.
w..Effective Sunshape The reflected beam profile can be characterized by using cone optics [24]. is usually small and on the order of 0. 1 where I = Idw. The incident energy on a flat differential reflector of area dA is. as shown in Figure 3. The reflection of radiation from a finite source.7: d2 e.5.5). from the central ray is .31 3. the magnitude of the intensity along an incident ray becomes negligible if the deviation. (3. .010 radian.7. to be: 27T f f d 2 e. smallangle approxi mations are valid. The irradiance on the reflecting element is found by integrating over the complete solar disc. (3. = sinC d dy). such as the sun. the scattering parameter asu n which is the standard deviation of the scattered distribution.1 = solid angle containing incident radiation (dwi In Eq.cose dA (3. therefore. is illustrated in Figure 3. = 2 1 = IbcosedA The total specularly reflected radiation is assigned the symbol e r and is composed of all the energy reflected to a surface At the normal to the specularlyreflected central ray.6) = Intensity distribution of the incident beam: assumed to be a normally scattered (Gaussian) function (Eq.5)).2 Reflected Beam Profile. reflector.(Cd~dy)cosdA 00 1 0 0 e.
7 Abstract representation of scattering by reflection. .32 o Incident Ray * Reflected Ray Figure 3.
Yy) such that 2= (Xx)2 + (Yy) 2 . Yy)I(x. It is also assumed that the surface is a good specular reflector.g. Consequently.7) This yields an expression for the reflected energy as d 4 e r = p(Xx. d 2 e. therefore.y) is given by (Xx. An element of is assigned to be of higher because the reflected order than the incident energy.. Yy) (3. the reflected energy.8) . a rather general formulation of the reflectance function is given by d 4 e 2 r d 2 e. y)(dx dy)(dA cose)dX dY (3. 2 su n * A small multiple of sun .. Cartesian coordinates may be assumed to exist on the unit radius hemisphere centered at dA since the angular deviation of a reflected ray from the specular direction of the ray from (x. and K is yet undetermined constant. The reflectance of the surface is given by the ratio of the energy in the incident ray to that in the reflected ray. It is assumed that the reflection function is of the form 2 K. descriptive of the material.exp( r /2 2 Mir ) where r is the angular deviation from r the specular direction for each incident ray. energy arises from integration over the entire incident image. the magnitude of the intensity of a reflected ray will also vanish if the reflected ray from the direction of specular reflection for a given incident ray is large with respect to the scattering parameter amir.5 osun). 1 p(Xx.33 large with respect to suffices (e. d 4 er.
and the resulting expression is easily evaluated by the method of Fourier transforms as follows: d 2 er (X. ) without risk of error from the small angle assumptions.exp((Xx) 2 + (yy) 2 )/2i 2 r Mir This is a reasonable assumption since one expects the reflec tance to be even in the deviations (Xx) and (Yy) and to vanish strongly with increased deviation. mir over the entire The preceding integrals are to be evaluated incident image. If the reflection is isotropic. The right hand side of the reflected energy evaluation becomes the product of two convolutions.9) The reflectance p is assumed to be of the form p(Xx. Yy) = K. x 2 2 )exp( ( ))dx dX dY 2Tro 2 sun S 202 sun 202 mir (3.Ib f exp( 2 2'70y sun 0 x 2 202 K 2 )exp((Xx)dx sun 202' dX dY mir .I b j exp(. it is permissible to set the integration limits to be (. as implied by the single parameter amir' the total reflected energy is given by: d 2 e r(X.Y)  K. Yy)(dA cose)dX dY (3.(Yy)2)dy(dAcose) S 2o 2 sun 202.10) f exp( _y2 )exp(. and since the incident intensity vanishes strongly for increasing values of X and Y.Y) K.34 Which is in terms of the reflected intensity: d4e r = Ir(Xx.
11) .Ib K. r sun Mir A similar procedure results in the same function for g(X) which allows the final representation of the reflected energy as: d 2 e 2 KI (dAcose) n2 sn sun (as )(X exp( 2 2 + y2) dXd( . 202.35 f _O2 exp( Y )exp( (y_ . )2 )dy(dAcose) 202 sun mir K.Y) = sun mir 2 +o mir ) ( +0 2 2(2 su +2. +G2. "i (3. the Fourier transform of the first product is readily expressed as F(a) = F(a) = F(f 1 (x) 0 f2 (x)) = esunmirexp( f(2 + su ir2 sun 2.)) mir ) This standard form can be evaluated in the X domain as: f(X) f()= sun Mir a2 %i mir exp( 2(02 x2 _  ) ) ) sun +02.Ib (fl1x)af 2(x))(gl (Y)@g2 ( y ) ) sun Since the Fourier transform of a convolution is the product of the transforms of the two functions.I b 0 b f 1 (x)f 2 (Xx)dx f g 1 (y)g 2 (Yy)dy(dAcose) 2 rC2 0_ sun 2 .
Y)dX dY = K. sun Mir )(dAcose)(dXdY) ) mir .. ) sun mir 2 ex exp( ( X +y2 ) 2(02 +G2 sun mir) (3. I d 2 e (X. (3. Y) = 27r(G +G2. bmir a2 R K 2 R.13a) . R. the reflected image has the radiant intensity distribution Rb Ieff(X. which is assumed to be measured by a wide aperture device normal to the spectrallyreflected central ray er =K.Y) as follows r R.36 The total reflected energy is then er f d2er(X.. YY) R 0 2 exp( (Xx) 2 2 Mir 2 0Mir )exp( (y_y) 2 / R exp( _ _2 2 2 mir 2 mir 202 mir And the reflected image in terms of intensity can be exprilssed by regrouping the expression for d 2 e (X. ) exp( 2(02 sun +02.Iba ir This allows the evaluation of K from the definition of the total spectral reflectance.12) The preceding expression is of the form d 2 r (XY)= IeffdwidA e Consequently.I *b mir The reflectance function becomes P(Xx.Y) = 27 (G 2 2 (X + y 2 ) b +G2.
being measured from the center of the solar For line focus systems like a trough it is convenient to transform the radial distribution to a linear one since a onedimensional description is required. Reference [25] presents empirical results which demonstrate that normal scattering is descrip tive of most actual surfaces encountered in solar energy utilization. disc.14) where is the angular coordinate in the plane normal to the angular coordinate parallel to the the focal line and .ned.37 or R.13b) tot where a2 r2 + 0 tot sun mir 2 + y2 2 =X The preceding results demonstrate that the reflected inten sity has the same functional form as the incident intensity when a normallyscattered source and normallyscattering reflectance is ass . The preceding derivation is based on a twodimensional sunshape description because solar intensity distribution data is usually reported as a radial distribution Iradial(0 in W/m 2 sr.I ieff(O ) a2 ) = b exp(tot 2 (3. The appropriate description is formed by simply assuming that lateral slices are taken across the twodimensional sunshape such that Ilinear = I radial ((C 2i (3.
(3. the one dimensional rms width (asunlinear) is related to asun (radial) by: (3.38 focal line (similar to the coordinates 01 Figure 3.1).2O2C2 7 tot where 2 2 +C2 tot sun.linear mir.linear( ) b ex( ea(2T) exp(22 b sun.linear 3.linear 2 sun.14). Similarly.linear Since the sunshape has a circular geometry. and e1. the sun appears to be narrower for a line focus collector than for a point focus collector where a two dimensional sunshape description is required. shown in Integration of Eq.13b) is modified as follows _ IbR  ) ) tot (3. results in an expression for the onedimensional solar intensity distribution: Ii Isun.16) Ieffa S=exp(. the reflected beam profile given by Eq.(3.5). using Eq.5.(3.15) = _linear In this sense.3 Analysis of Optical Errors In this section models are presented for characterizing and assessing the effect of optical errors described in section .
that is. are assumed to be approximately normal (Gaussian) distributions with zero mean [16. parabola profile errors. a statistical approach for modeling the effect of the errors is described. receiver location errors. . when averaged over time and over the entire collector or array of collectors. all five errors are assumed to be statistically independent stochastic processes and their occurances. 22]. reflector alignment and tracking errors. local slope errors. a normal distribution function remains a normal distribution function when convoluted with other normal distribution functions. the resulting scattered intensity distribution of the energy directed to the absorber tube (from the reflector). Then. is obtained from the convolution of the sunshape successively with the normal scattering functions describing the independent modes of scattering. Statistical Error Analysis In the statistical error analysis.39 First. As shown in the preceding section. the independent optical errors are: " " " " " mirror diffusivity. Once again. Then the shortcomings and errors associated with this approach are discussed and a comprehen sive approach for modeling the effect of errors (on the optical performance of the trough) is presented. the effective sunshape.
40 Therefore. Obviously. Bendt [16] presents 0 tot = /0 2 + G2 + (2co)2 + ac + a 2 sun mir slope )a displacement (3. a normal (Gaussian) distribution approximation for the optical errors permits the convolution calculations to reduce to simple addition of the standard deviations a and allows for the characterization of the errors with a single error parameter.5.17) where. the tracking error distribution (timeaveraged over several oscillations of tracking error for an entire field of collectors) and the receiver location error distribution (averaged over an entire field collectors and time). the effective sunshape is a normal distribution with an overall scattering parameter. the described statistical approach has a lot of merit. the slope error distribution (averaged over an entire collector surface). the mirror nonspecularity error distribution. atot as follows: For example. shown as in Figure 3. terms under the radical represent the standard deviations of: the sun's energy distribution. atot' equal to the square root of the sum of the squares of the individual scattering parameters. it provides a 'commondenominator' for the combination . Twice the standard deviation of the slope error is employed because. In sunimary. reflector optics result in doubling of the slope error as the deviation from the mean specular direction.
errors by normal distributions with zero mean is not valid when small scale applications. i. However. and over a long Similarly. reflector alignment and tracking errors can be assumed to be random (and. have normal distributions) only when they are averaged over an entire field of collectors time. In a recent study. representation of these to the same restraint. even if all of the errors are forced or assumed to have normal distributions. Furthermore.. which is the backbone of this approach. aot' obscures most of the information a designer will need during the preliminary design phase. small installations for home heating. Therefore. consequently. are analyzed. for example.41 of various errors into a single parameter (atot) . For example. Treadwell [26] addresses a similar 'gross' optical issue and attempts to quantify the effect of tracking and receiver location errors separately. it is necessary to handle some of the errors differently and have more than one error parameter to characterize and quantify the geometric effects resulting from error distributions that do not have zero mean ( This will also help incorporate the effects of errors).e. and thus simplifies the analysis and calculations tremendously by reducing the number of independent variables. is not valid for all errors for all practical applications. the receiver location errors are subject Therefore. separate from . the normal distribution assumption for errors. the resulting single error parameter.
errors are divided into two Then. error parameters are obtained for the characterization and analysis of the different types of errors. therefore. a more general and comprehensive approach is taken.) Moreover. groups: First. the analysis of e. he fails to introduce new 'universal' error parameters.rors are embedded in a complex collector model which takes into account geographic effects as well. in his study. Comprehensive Error Analysis In the present study. universal random and nonrandom (systematic). There fore. are treated statistically and give rise to spreading They .42 the effect of the rest of the errors that are lumped into atot" However. and as a result. ends up presenting results for a limited number of cases. can be represented by normal (probability) distributions with zero mean. Random and nonrandom (systematic) error groups are explained as follows: Random Errors Random errors are defined as those errors which are truly random in nature and. (Three concentration ratios and two reflected energy distribution standard deviations are considered for a 90 0 rim angle reflector. the presented results are highly restricted and lack the degree of flexibility a designer will need when designing troughs for different environments and sites. thereby making the results sitedependent.
[17] have recently shown that scattering that results from slope errors can be measured using laser raytrace techniques and the value of the scattering parameter. [29] without Bendt present methods for determining aslope laser ray tracing. p.43 (widening) of the reflected energy distribution (see Figure 3. plastic and metals. waviness of the reflector surface due to distortions that may occur during manufacture and/or assembly). et.g. Reference [25] presents empirical results which demonstrate that normal scattering is descriptive of most actual surfaces encountered in solar energy utilization. be characterized by normal scattering Pettit et. Slope errors of this kind can [28]. These errors are identified as follows: (i) Scattering effects that are associated with the optical material used in the reflector. (ii) Scattering effects caused by random slope errors (e. Reference [27] lists values of specular reflectance.2) in the construction of the reflector.al. (iii) Misalignment of the collector (with the sun) due to purely random tracking errors (which last only .8). and scattering parameter amir for glass.. al. a slope' can be related to the fabrication method used (see Table 3.
=oMean = 0. . 3 = Misalignment and Tracking Errors Concentrator Figure 3.44 Sunshape Normal Intensity Distribution . 0 Glazing Receiver Tube with0 Stand.8 Modeling of potential optical errors in parabolic trough collectors. Dev.flatter distribution due to random errors.0. /LIdeal Central Ray Mean . = totn * . Dev.0 Actual Ceitral Ray / Incidentay Centl Ry < Normal distribution with Stand. Dev.R Effective Sunshape (. = c'totn _mean = . shift of central ray due to nonrandom errors.) Normal Distribution with Stand.0 .
8 77. .9 (cm) (mrad) 7. 38 = thick Bent glass over Foam Glass High Pressure Laminating Press Low Pressure Press 73. 51 m thick Aluminum Skin and 9..3 2. 2 1976 Dollars.9 74.3 Laminates Plywood 19 = thick7 ply Fiberglass Skin Corrugated Fiberglass Core.5 m Cell.4 (ca) (=rad) 7._ From Reference [17].5 2.9 2.9 2. 72.0 1.1 80.0 1.5 77.1 77. a After 13 Months o Temperature and Humidity Cycling Focal Length RIS Slope Error.8 76.2 ca.5 Cin Low Pressure Press Autoclave 1. 19 m deep Note the nominal focal length was 76. Aluminum Honeycomb Core.2 4. a (8/rt ) l.2 Typical Focal Lengths and Slope Errors for Model Parabolas of Various Materials and Processes As Manuractured Types of Structure Materials Fabrication Method Cost Estimates 2 2 Focal Length FINS Slope Error.T Vacuum Forming Closed Die Forming in Hot Hold 3.5 78.0 Sandwiches 2.I TAEIE 3.4 In Testing In Testing Molded Structures Sheet Holding Compound Molded with Integral Ribs.6 4.5 2.5 74. 51 m thick Paper Skin and Corrugated Paper Core.2 1.
as those errors which can be related directly to anticipated errors in manufacture/assembly and/or in operation. change in the location of the focus of the reflector. due for example to a constant tracking error or rotation of the trough's vertextofocus axis during assembly. in general. due for example to deflection (flattening) or severe waviness of the They will cause a permanent reflector surface. Hence the reflected rays may miss the receiver.8). 'non They are characterized.46 a very short period of time) can also be assumed to have normal distributions. In general these errors will cause the central ray of the reflected energy distribution (effective sunshspe) to shift from the design The nonrandom errors are direction PO (see Figure 3. instead of a distribution of values. are called random'. This will shift the location of the focus to the left or to the right of the ideal focus and may cause the central ray of the reflected . (ii) Consistent misalignment of the trough with the sun. if the receiver is located at the design focus of the parabola (reflector). Nonrandom (Systematic) Errors Errors which are assumed to have a single deterministic value. identified as follows: (i) Reflector profile errors.
sun's incidence and solar the .19) distribution sun.n + 2 4C slope. (3. = Angular aperture (radians). by a total reflected energy distribution standard deviation at normal incidence. 2 Ib = Beam solar radiation. W/m = Angular shift of the mean of the distribution (radians).18) = teff( ot. The scattering parameter for random errors is given by: atot. mirror. In the present study. p = Average specular reflectance of the reflective surface.n = Energy rays at normal standard deviation of noon.n + .3.n + a 2 .n where.47 1 beam to miss the receiver. random errors are modeled statis tically. This kind of error will also be systematic and it will cause the central ray(s) to miss the receiver. (iii) Misalignment of the receiver with the effective focus of the trough (due to reasons discussed in section 3.1).n= =/y2 where. a sun. )2 ) (3. atotn and the effective sunshape is expressed as: Ion(b PI b exp( liJ . 2 tot.
e.n = diffusivity of the reflective material at normalincidence.. they can be accounted for together. " Misalignment errors : Angle between the central ray from the sun and the normal to reflector's aperture plane (B in Fig. The nonrandom errors are analyzed as geometric effects. in a single parameter. Because reflector profile errors and receiver location errors along the optical axis (in ± y direction) essentially bring about the same effect (see Figure 3.3). The following physical measures will be used in quantifying the nonrandom errors: " Reflector profile errors : Distance between the effective and ideal focus of the reflector measured along the optical axis of the reflector.48 Cslope. Cmirror. " Receiver location errors : Distance between the ideal focus of the reflector and center of the absorber tube. are Hence. provisions are built into the optical model Each one which allow for the analysis of such errors.3.9). n  standard deviation of the distribution of local slope errors at normalincidence. there only two independent nonrandom variables and these (Note that dr is the distance between the are a and dr• effective focal point of the trough and the center of the . of these errors will be characterized by a single determi nistic value. i.
I (d r )y =receiver location error perfect parabola (a) Receiver location error Eftective focus of deflected parabola Perfectly located receiver at ideal focus . y .49 Ideal focus Dislocated receiver . . 4 (d r.9 Combination of receiver location and parabola profile errors.reflector profile error perfect parabola deflected parabola (b) Parabola profile error Figure 3...
the three error parameters will allow for a comparison of the effects of different errors and help to determine and minimize the 'crucial' Also. Moreover. errors (errors which degrade the performance the most).) The nonrandom errors are modeled as geomet ric effects and they are accounted for by a finite angular shift in the mean (P in Eq. (ii) In croughs of finite length there will be loss of radiation from the ends of the reflectors (end effects).50 absorber tube. as it will be seen in the following chapters. this will allow for a tradeoff study between errors.1) has (see an effect on the width of the image on the receiver (abberation effects). This will. Offaxis operation affects the performance of the trough in two ways: (i) The elevation of the sun from the xy plane Figure 3. having three error parameters will substantially enhance the effectiveness and reliability of design calculations. complicate the optical analysis. In summary. n ) and two nonrandom (a and d r).6 Effect of OffAxis Qperation The collector is said to be in offaxis operation when the direction of the incident central ray is not perpendicular to the aperture plane of the trough.18)) of the reflected energy distribution. to characterize various optical errors. . of course. one random (a tot. instead of one now there are three error parameters. 3. However.(3.
all the offaxis effects including the variations in the width of the solar image are assumed to be lumped into an incidence angle modifier. [16] have shown that the width of the solar 1/cos. which is to be experimentally measured as recommended by the researchers at NASALewis and Sandia Laboratories Rabl et.[18] [31].al. It is assumed that this effect will be accounted for when the decision on the length of the ATstring is made after the application type and the tracking mode are known. variations in the width of the solar image are not taken into account explicitly.. therefore.17) will need to be changed to: _ sun . n a cos 11 asun (3. long troughs or polar mounts [30]. . Instead. have presented empirical equations for experimentally measured incidence angle modifiers for several parabolic trough collectors manufactured by various U. for offaxis operation. The end effects are not taken into account in the present collector design studies. in the present work. the end effects can be minimized by using end reflectors. Hexel (SERI) and Del.20) However. . Solar Kinetics.51 Nevertheless.al.(3. image on the receiver will vary as Therefore. Rabl et.S. K(O). Hexel (Sandia). the scattering parameter a sun in Eq. design studies are based on instantaneous onaxis operation only. companies like Acurex.
7 Summary In this chapter a thorough study of the optics of the PTC's was presented. is a function of collector geometry parameters C. of the collector on performance was also discussed. p and D as well as the random (a) and the non The effect of offaxis operation In the random (3. for modeling and analysis of the errors. . Also. the derivation of universal error parameters which combine collector geometry para meters with error parameters is given. intercept factor. preliminary results of the comprehensive optical analysis are presented. following chapter. dr) error parameters. fraction of the solar rays intercepted by the collector absorber tube. assembly Models were developed It was shown that the and operation of troughs were identified. Potential optical errors in manufacture.52 3.
Also. As indicated earlier. dr) 6 (4. 15. the optical errors as random assuming normal distributions (with zero mean) for their occurances. The code has been modified to handle random and nonrandom errors. onto circular absorber a tube and then to output. of the EDEP code [32] A modified version developed at Sandia is utilized for analyzing the effect of errors on the intercept factor. y. can be written as: y = fn(P. the intercept factor at normalincidence. given by Eq. these were modeled as geometric effects. 26].1) where D is the absorber tube diameter. instead of starting from scratch.n' . C. So. This code uses a raytrace technique to project the effective sunshape. The EDEP code was used extensively in design and analysis of Sandia's prototype troughs [13.16).Chapter Four PRELIMINARY OPTICAL RESULTS AND DERIVATION OF UNIVERSAL ERROR PARAMETERS The value of the intercept factor is governed by random and nonrandom errors as well as the concentrator geometry (concentration ratio C and rim angle $). D. It has been validated and shown to be sufficiently detailed a optical simulation model. the amount of energy intercepted by the absorber tube. Provisions were built into the code to handle deterministic angular pointing errors (a) and receiver mislocation errors (dr). (3. the original code handled all. 14. the code has been restructured in 53 . a = atot. Therefore. the decision has been made to utilize EDEP's basic numerical structure as the starting point in the present study.
the simultaneous effect of receiver mislocation in the lateral direction.3 are unsymmetric. D=2.54 order to facilitate its use and applicability to comprehensive design studies and determination of errortolerances. receiver mislocation .(3. a. expected. dr) for a fixed col lector geometry (C=20. This is expected because tracking errors move the focus of the the reflector along the lateral axis and the effect of this shift is either compensated or magnified by the lateral receiver mislocation : for a positive a.1.54cm and 4=90') is studied. is chosen for the random error parameter. These are accounted for by a separate multiplicative factor. C and D) and error parameters (a. 4.8.18.1)). Results of these runs are summarized in Figures In Figures 4. which is the standard deviation of the energy distribution of'the sun's rays at solar noon on a clear day.1 through 4. lower limit of 0.1).0025 radians. and tracking errors are studied for different A values of random error levels (Figures 4.1 to 4. curves in Figures 4. PTa. effect of changing random and nonrandom error parameters (0. a and dr. 4. No transmissionabsorption losses are included in the geometric intercept" factor (Eq. as shown in Eq. First. (4. Many computer simulations were made using different sets of values for the parameters p. code is named EDEP2.3). (dr)x.1 through 4. 6.2 and 4.B and dr) and it outputs the geometric intercept factor. As receiver mislocation in the lateral direction produces a non uniform effect when combined with simultaneous tracking errors. C. a. The new Its input includes geometry parameters ( .
0 a 0.5 D2. / o0. .0025 radian.0 a a 0.7.2 0.3 __0.1.0 I g 1.1 Intercept factor as a function of receiver mislocation along the lateral axis and tracking error for a = 0.0025 rad] 0.8 a .cm Figure 4.8 I I 1.4 a a 0.   1..2 I I 0.8 * 0.2 1 0.4 .0 0./ 0.5 0. I 0.9 .4/ / I'  a I/ / 0.3 I 0.6 0.6 0.0 I 1 1.0 0.2 1.6 0.(dr)x.4 0.7 0.0 Receiver Mislocation .54 cm 0..9 /0. 1I 0."" 1 0.8"'  .
8 1.1.0 1.0.6 0.5 .2 0.0 0.3 0.5 "C 2 0 D2.2 0.Cm Figure 4.4 0..0 1.2 0.6 0.54 cm l0 ..6 0.0 1..4 0.2 1.0 0. 1..3 0.. .5 0.2 0.4 0..6 0. _.0 0. (dr )..0 0 .4 0.4 0.9 .01 radian.2. 0.8 .2 0. ..2 Intercept factor as a function of receiver mislocation along the lateral axis and tracking error for a = 0.7  0..9 0.8 1. 00.4/ r0o.7  0.0 0.8 0.0 Receiver Mislocation.2 1.01.
0.2 0.7 .8 I I 0. .0 0.1. 0.4 0. .3 a. 0.2 .02 rad 0. D.6 0.2 I .02 radian.6 .5 0 04 Cm 20 0.0 1.4 0.0 i 1.9 . 0.2.0 I . 1.8 0 U~c 0. .3 0. I 0.4 .6 0. .0 1.2 1. 1.0 ' 0. 0.4 0.. (dr)x.6 . o.0 . 1. I 0.8 . cm Figure 4.0 0. .6 0.9 0.0 . .0. Receiver Mislocation. 0.4 0.0 I . . 0.2.2 0.54cm 0.8 1.5 5:.0 0. .2 1.2 . 0.8 0.3 Intercept factor as a function of receiver mislocation along the lateral axis and tracking error for o = 0.80 C)~ .6 .
i. * Reflector profile errors shift the effective focus along the optical axis and therefore.e. (dr)y . the axial component of dr.58 mislocation in +x direction compensates for the tracking error while mislocation in x direction compounds the effect of the tracking errors. which . higher random error levels the curves are flatter. curves in the figures are symmetric. ±(dr)y location errors. this will eliminat the complications that may arise due. the simultaneous effect of receiver mislocation along the optical axis (vertextofocus axis) and tracking errors are studied for different random error levels (see Figures 4. For the design studies.4 through 4. Again. receiver mislocations along the . in As shown in the fig ures. this is due to flattening of the intensity distribution of the effective sunshape. Next. therefore. In this study. it will be appropriate to consider receiver location errors along one axis alone.6). for Obviously. for a receiver located at a distance dr away from the effective focus. receiver mislocations +y direction bring about the same effect as the mislocations in y direction. as expected. arguments: are chosen to represent the receiver This coice is supported by the following In most reallife applications occurance of receiver location errors in the ydirection have the highest expectancy because the sagging of the receiver due to its own weight and/or due to thermal expansions will result in mislocation in this direction.. optical axis. to odd coupling of the errors. for example.
4 0.9 '.0 1.0 0.0 0.0 0.7 >.32./ 0.2 .4 0.8 1.0 1.3 // 0.6 0.0 1. \0.4 cm___ _ 0.4 Intercept factor as a function of receiver mislocation along the optical axis and tracking error for o = 0. D7.4 0.0025 rad Y0.5 100.8 /0 o.2 0. 0.0025 radian.8 .8 1.0 0.0 0.5 0.8 IN0 2) 0.6 0.3  Cr .0 Receiver Mislocation. 0.7 0.2 1.2 0.(dr)y.2 0.0. . cm Figure 4.2 1.6 0.6 0.9 \.4 0.4c C20 0..2 0.3 0.4 0.2 1.8 0. 0.0.0 1.6 0.1. 0 0.
2 1.2 0.4 .  0. .5 0.54 cm + y 0.8 0.0 0.0.2 1.5 Intercept factor as a function of receiver mislocation along the optical axis and tracking error for a = 0.0 1.6 0.0 1 1.6 0.2 0.4  C20 0 "dr .9 o •0.2.8 I 0.0 0.4 0.. 1.0 0.2 0.3 0.2 0.. Ca .4 0.2 1.0 0.6 0.0 o.0 0.8 0 L .0 0. 0.5 0.II.4 0.4 0.8 1.3 0.0 Receiver Mislocation.6 0.011r~] 1. 0.(dr)y.01 radian.1.6 0. cm Figure 4.8 1.2 0.
1.6 1 1. o 0 9 0 dr 0.3 0.500 0.012 1.8 .000 0. .2 0.8 0.4 0..(dr)y.0 0..3 o 0 0.0 0.6 0. 1.250 0.750 0.9 S0.6 0.7LA. L .6 Intercept factor as a function of receiver mislocation along the optical axis and tracking error for a = 0.0 3.0 0.2 0.8 0. .0 .8 1.54 cm.2 '0.2 1.02 rad ] y 0.1.0 0.5 +y .7 0 ./0.8 1.4 0.. 0..2 Receiver Mislocation.5 0.8 0.0.0 1.0 1. cm Figure 4.8 0.2 0.0 1.4 0.6 0.4 0.0 0.2.6 £0.40.2 0.00 0.4 0.9 /.0 .. o s C20 D.02 radian.
Unsymmetric effects produced by the receiver mislo cation in the lateral axis will introduce undesirable complications into the generalization of the results.3 with Figures 4. figures are presented justifications for combining the error variables a.62 combines reflector profile errors with the receiver mislocation errors along the optical axis (see Figure 3. y. Therefore. In Figure 4. (dr)y can be taken as the dominant direction for receiver location errors. C and D is studied. +a.8). .18. for different random error levels. and (dr)y with the geometric variables C and D to yield a set of three error parameters universal to all collector geometries.6 shows that. when receiver mislocations in the ydirection alone are considered. for a given tracking error. positive and negative tracking errors produce the same effect and of course this will be helpful in the generaliza tion of the results that will follow shortly.7 and 4. the effect of positive and negative tracking errors (± B) for different random error levels is studied (Figures 4. B.14. the effect of changing Also in these the geometric parameters. will be larger than lateral component of d (dr )x Comparison of Figures 4. and receiver mislocation in x direction exist together).9 through 4.9). In Figures 4. the intercept factor drops more rapidly when the receiver is mislo cated in the ydirection (except when positive tracking errors. Next.44. As shown in the figures.9 concentration ratio C is plotted versus The intercept factor.
..0 .25 1.0 1.75 1.5 0.deg Intercept factor as a function of tracking error and random errors for (dr)y = 0.75 1.5 0.25 0.0025 rad0 0 0.0 0.0 0.1.0 0.0. Figure 4.25 0.25 0.75 0.5 0.7 /3 .5 00. cm (dr)y 0.0 1. 0.4 .d.50 Il 0 1.75 0.0 1 1.5 OI 0/ 0.25 Concentrator Misalignment and Tracking Errors.0 and 0.5 0.4 0.0 0.25 0.508 cm.
9 0..25 1.. 0.0 1 1.80 LL 4wd .8 for (dr)y = Intercept factor as a function of tracking error and random errors 1.25 0.8 0 0.5 0.deg 1.7 . 1.5 0. /3 Figure 4...25 Concentrator Misalignment and Tracking Errors.25 0.6 C02 0 0..424 cm.0 0.0 0.25 0.5 0.0 0.75 1  1.0 0.6 0. 02 0.75 0. .0125 rad 0. 0.016 and 1. L (r o.0075 rad 0.5 0.25 0.4  C)..0175 rad 0.. ..9 0.75 .1.0125 rad 0.0025 rad 0...0 0.7  .0 1..424 cm 1 0.016i 0.0176 rad 0.0225 rad 2( 1 1 1 1 u.0 0..05rdD25 m.75 0.0 1.0 1 1. .
3 0.0 75 100 0 25 50 Concentration Ratio.0 0.0 0.8 0. o 0 .2 13= 0.0 5 rd 0. 2.0 0. C Figure 4.9 .54.6 0. 5.1.08 cm 0 0"'25n" 0.2(dr)y = 0.1 0.6 CL U1) 0.5 0.000" = 0.0 I I 0.27.8 0.00 1. .7 0.4  D 1.9 Intercept factor as a function of concentration ratio and random error parameter (zero nonrandom errors).
66
results are repeated for three different values of absorber
tube diameter and it was found that, for perfect tracking
(U=0)and perfectly aligned receiver (i.e., (dr )y=0), the
intercept factor is not a function of the absorber tube diam eter. As shown in the figure, the intercept factor decreases
This decrease is more
In Figure 4.10,
as the concentration ratio increases.
pronounced for higher random error levels.
the intercept factor is plotted against a new parameter,
,
a =aC, which is called the
'universal random error parameter'.
This parameter collapses the five curves in Figure 4.10 into a
single curve.
In Figures 4.11 to 4.13, again the concentration ratio C is plotted versus intercept factor , this time for different
tracking errors and for three fixed random error level,
,
a =0.05, 0.1, 0.2 rad.
As expected, the figures show that,
intercept factor rapidly drops to zero when concentration
ratio increases; this is particularly true for large tracking
errors. Again, it is found that, for a perfectly aligned
receiver, i.e.,
(dr)y=0.0, intercept factor is not a function
Figure 4.15 shows that curves
of the absorber tube diameter.
in each of the Figures 4.11, 4.12 and 4.13 reduce to a single curve
when intercept factor is plotted against
,
a 'universal nonrandom
error parameter',
6 , which is the product of the tracking
error and the concentration ratio. Figures 4.15 and 4.16 show plots of intercept factor
versus concentration ratio for different receiver location
1.0
1.0 1
0 o 0.0025 rad 0 0.0075 rad
1.0
0.8S
A
0
0.0125 rad 0.0175 rad
0.8
0
C3 L C1.
0
0.0225 rad
0.6
0.6
o
'
0.4
D = 1.27, 2.54.
0.4
0.2
0.2 5.08
= goo
cm
0.2
(dj)y = 0.o0
0.0/ = 0.00. 1.0 2.0
Universal Random Error Parameter (o* = oCW,
rad
Figure 4.10
Intercept factor as a function of the universal random
error parameter (zero nonrandoin errors).
1.0 0.9 0.8 o
LL
0.0"
1.0
A9.?
0.8
0.7
CU
0.6 0.5 0.4 0.3 0.2 
0.1
L
0.05O
rad' .Jo
0.6 0.4
C.
0

D = 1. 27, 2.54.
5.08 cm I
0
90,
(dr)yC.O
0.2
0.0 0
1 25
50 Concentration Ratio, C
75
0.0 100
Figure
4.11 Intercept factor as a function of concentration ratio and tracking
error for o* = 0.05 radian and (dr)y = 0.0.
8 0 0.0 0. 10 rad U.0 "€* 0.0.10 radian and (dr)y = 0.27. 0. Cm 2. C Figure 4.08 D .54.01.1.0.8 0.6  o* 0.o01.4 ~5. (drly .1.0 k =o.6 0. . 1.4 00 O0 215 50 75 IO Concentration Ratio.12 Intercept factor as a function of concentration ratio and tracking error for a* = 0.
0 /= 0.0.13 Intercept factor as a function of concentration ratio and tracking error for o* = 0.1. 0.0 : 0.08 cm go* ) ° (dr) v x 0.2 0 0.4 5.54.6 0.8  0 0. 0.00 1.0 100 Figure 4.8 0.4 D . 0 L 0. C 75 0.6 o U. 0. .1.27.2 0.20 radian and (dr)y = 0. 2.0 0 1 25 50 Concentration Ratio.0  0.20 rad 0.
2 D = 1.0 ....  0.250 1.4 0...1.. 0 0! 0 0.. 1.. C 0 1 /3=0.8 )k.2 .....0 0..500 0. 2 CDao 0...... 5..20 0.0. 0.6 0' C 0.05 rad M 0...08 c (dr)y " 0.....14 Intercept factor as function of the universal nonrandom a error parameter due to angular errors (0*) and various random error levels (zero receiver location error).. 0..4 \..54. 0... 2.0 ....0 Universal NonRandom Error Parameter (/3"=/3C)..8 0 0 0.. ".. ....000 0. rad = 0.....750 1..00 0..10 rad = ...6 0..27.. rad Figure 4......0 L..0 0.
254 cm 0.0 50 Concentration Ratio.o cm 0.0.27 cm (dr)y 1.0.27 cm 0.27 cm. 0 D 1.15 Intercept factor as a function of concentration ratio and receiver mislocation for D = 1.3175 cm (dr)y="O.127 cm 0. .oo 0.0 > L0 0.0 (dr)y "o.2 .6 0.8 00.835 cm l 40 =1.1 rad .8 0.01 0 0. 0.1 rad I 25 I0J. C 75 100 Figure 4.0.2 0.1.4 0 .6 0.
8 L 1.3175cm 0.635 cm (dir).16 Intercept factor as a function of con.0 (dr).0 0. .127 cmi.0. C 75 100 Figure 4.6 .27 cm 0.0 Cm 0.254 cm 0.1 (dr)  2.4 D2.2 0_90* o'0.entration mislocation for D = I ratio and receiver 2.i 0. 1.8 0 U o0.1. 0. 0.54 cm.54 cm 0.i rad I i 0. o O.0 25 50 Concentration Ratio.54 cm 0.2 red 3o0.6 J.0 0.
8 1 cm 0.08 3.6 0. D.0 1. cm Figure 4.8 0 (dr)y 0.1 rad 0.0 5.2 a=0.(dr)y= 0"0 0. rad 10. .17 Intercept factor as function of receiver diameter and a receiver location error.1.4 C:: (dr )y =3.81 Receiver Diameter.3175 cm1.635cm 0.8 0 cc00.0 y(d ry = 0.2 0. 0.0 ' .4254c 0.6 2 0 C1 .54 0.27 2.
0 1.8 0. .0 I 0.0 0.2 0.5 1.0 1.6 0.0 1.0 0.1 rad p 0.2  o =0.18 Intercept factor as a function of the universal nonrandom error parameter due to receiver mislocation.0 0.0.4 0.5 Universal NonRandom Error Parameter (d'= (dr)y /D) Figure 4.60. 0.5 1.4 0.8 S LL 0.1 rad 3 = 0.
((dr)= distance between effective focus and ry center of absorber tube measured along the axis of the reflector profile (see Fig. a = aC = Universal random error parameter.2) where.76 errors. to yield the second universal nonrandom error parameter. Next. d ) (4. have been reduced Therefore. It is evident that for fixed values of a and . As a result of the discussed compater simulations. (c = standard deviation of the reflected energy distribution) [rad] = BC = Universal nonrandom error parameter due to angular errors.17. d =(dr ) y/D. shows the combination of the receiver location error.18 (dr)y . This reduction in the number of independent variables in (Eq.1) tc four. the equation can be rewritten as: * 6* . a . (a = tracking or reflector misalignment error) [rad] d*= r D = Universal nonrandoni error parameter due to receiver mislocation. (4. with the absorber tube diameter. 3. Finally. the number of independent variables in 7q. * y = fn(p. this plot is shown in Figure 4. the inlercept factor is not a function of C.1) is a significant finding because it allows for gener . (4. the intercept factor is lotted versus absorber tube diameter.9)) [dimensionless]. Figure 4.
77 alization of results and meaningful optimization of the collector geometry. The next chapter parameters are presented elsewhere presents results of optical analysis using the universal error parameters. Analytical derivation of the universal [33]. .
The errortolerant trough design concept is particularly suitable for developing (ountry appica tions. were presented.Chapter Five RESULTS OF THE COMPREHENSIVE OPTICAL ANALYSIS 5.1 Introduction In the previous chapter. In particular. Finally. Chapter Four. the sensitivity of the error tolerances to changes 78 . 'eriortolerant' trough design (i. the effect of changing The concept of designs which have built the collector rim angle is studied. a set of universal error paramters which combine the collector geometry parame ters with the error parameters were derived. the effect of the combined random and nonrandom errors are studied. Second. the effect of the random errors alone cn the intercept factor is analyzed. assembly and oper ation errors) is introduced and methods for determination of error tolerances are given. in tolerances for potential manufacture. the results of an indepth optical analysis of PTC's (using the derived universal error parameters) are presented. First. the preliminary results from the comprehensive optical mode. In this chapter.e.. As a Yesult of the preliminary analyses. Results from this analysis are compared with those of the previous investigators.
lector for a less than 0. a ) With this limitation the results can be presented in two dimensional graphs. 5. angle reflector performs better than 120*rim angle re..0) Eq. a 105'rim with higher rim angles.1. itself after a rim angle of 1050.e.2 the optimum rim angle based on the random errors alone. the trend starts to reverse For example. shown in Figure 5. Figure 5.2 versus rim angle for different random error levels.1 shows a plot of y versus a shows the cross plot. toabsorber distance oZ the higher rim angle reflector. (4.. is so broad (in the range of 800 to 1200) tbft the final choice of rim angle can be determined by other con siderations as: (i) (i.2) becomes a function of rim angle and the 'universal random error parameter' only. higher random error levels can be tolerated However. and ease of manufacture.e.79 in rim angle (i. The results presented in this section can be compared .. y = fn(p. y As shown for different rim angles while 5.0 and d*=O. optimal rim angles) is studied.i) (iii) nonrandom errortolerances. i. mechanical strength of the reflector.e.2 Result Of Optical Analysis With Random Errors Alone For zero nonrandom errors (i.22). in Figure 5. 6 =0.2 radians (curves cross each other at This is due to the increase in the averae reflector As a =0.
6 0.609 0.4 0.8 0.0 0.8 0.00 1.1 Intercept factor as a function of the universal random error parameter and rim angle (zero nonrandom errors). rad Figure 5.80 0 0 u. 0 0120 0.0 1.0 __1.6 y:0.00 1 0. 00.80 0.2002 0.80 y= 0. 0.75 0.0 1.00 >.00 0.4 0. .2 0.2 .95 0.60 II 0.0 r'. 0.
45 60 75 90 105 120 Rim Angle.3 0..0 30 . locus of optimal rim angles for zero nonrandom errors 0.4 0. deg.030 45 60 75 90 105 120 0.1 1.2 Intercept factor versus rim angle and the universal random error parameter (zero nonrandom errors). 0 0 c0. .0 .7 L. Figure 5. I ...81 1.2 6 L 0....4..6 C1 0.80.
Figure 5. [16 and 34].13 show the allowable level of error The vertical in order to achieve an intercept factor of 0..95) for rim angles 80' to 1200.9 to 5.they did not explicitly report factor. (y=0. the two results are This close agreement validates The present results the present calculational procedure. .1 and 5. 0..3 Results Of Optical Analysis With Random And Nonrandom Errors Figure 5. in their studies [13.82 to those of the previous investigators who assumed normal distributions for all optical errors. Figures 5. Figures 5.95.75. the optical efficiency or the geometric intercept Instead their results display only the overall collector conditions which makes direct impossible. Then. efficiency for chosen operating comparison with the present results 5. for a given intercept factor level the nonrandom error effects can be shown in twodimensional curves.8 illustrate the allowable level (in order to achieve an Like of random and nonrandom errors intercept factor of 0.15]. could not be compared with those of any other investigators because. wise.75.2 were compared with similar plots found in Ref.). found to be almost identical. where a purely analytical technique is used to calculate the geometric intercept factor (opposed to the numerical ray trace technique used in the present study).4 to 5.3 shows that for a given random error level (a*) and rim angle ( ) the intercept factor can be plotted as a function of * and d in a threedimensional graph.
0.b a =0.15 Al Figure 5.75 . ..0) (90 .0.0..Y=0.0.0.l /J 7'= 0.0.3 Qualitative representation of intercept factor as a function of the nonrandom error parameters for fixed rim angle and various values of random error parameters.83 1= 9oi (900..0.95 .0.15.0) 7.0.
2 0.1 0.rad 0.95 0.5 .84 0.5 Y(80.3 0.* .4 0.1 0.2 R3.1 0.15 0.2 0.1 0.95 and 0 = 800 and various random error levels.. .d*)= 0.4 Figure 5. 0 0.2 0.3 0.4 0.3 0.2 0.1 0.4 Allowable nonrandom error levels for y = 0.21 G_9III I .3d" a =0.4 0. 0.
0 0 .4 Figure 5. 0.3 0.5 goioi 0.1 0.5 Allowable nonrandom error levels for y = 0.0 =0.2 0.4 d@ 0.95 and 0 = 900 and various random error levels.0 : 0.2 0._L_ 0.4 0.3 0.4 0.2. .rad 0.1 0.15 0.1 0.2 0.85 0.3*=0.1 0.2 0.5 0.2 /3*.3 0.
2 /Irad 0.2 0.86 o.15 0.4 1000.0.4 d 0.I 0.95 0.1 0.3 I 0.6 Allowable nonrandom error levels for y = 0.0 .1 0.2 0.1 0.1 0.4 0.3 0..3 =0.5 0.2 0. 0 0. .5 Y (10 0 .95 and = I00' and various random error levels.3 0.) = 0.2 0.4 Figure 5.1 0.
3 0.5 0.3 0.d) r 0.4 0.1 0.95 0.2 0. 5.1 0.2 0.3 0.rad 0.4 Figure.2 0.3d** 0.1 0. .510 y( 1 0 *.7 Allowable nonrandom error levels for y = 0.1  o.4 0.4 ILII 0.1 0.20.0' 0 0.87 0.15 0.95 and = 100 and various random error 1 levels.2 $* .
95 and p = 1200 and various random error levels.8 Allowable nonrandom error levels for y = 0.88 0.c.4 Figure 5.4 0.5 4)100.1 0.15 0.1 0.2 0.3 0.3 a* =0.5 (1200.2 0.1 0..0 I 0 0.2 0.3 0.2 "*.1 0. .4 0.95 0. d*).=0.2 0.R*.3 0.1 0.4 d1 0.rad 0.
2 'a .2 0. 0 *.5 0.30.3 I 0.3 0.2 0.0 0 _ _ 0.rad 0.4 Figure 5.4d0 0.9 Allowable nonrandom error levels for y = 0.1 0. o'" "0.1 0.3 0. .4 0.5 y (80.2 0.15 .1 0. d) 0.35 0.3 0.2 0.75 and 4 = 800 and various random error levels.75 0.89 0.4 0.1 S0.1 0.
1 .o*=0.3 0.90 0.415 0.2 /* .5 Z y( 90 o .3 0.4 0.5 0.1 I 0.1 .0 1 .10 Allowable nonrandom error levels for y = 0.1 0.d*) 0.4 O.0.2 0.1 0.2 0.* *.4 00.2 8d. .15 . 0.3 d* 0.3 0.25 0.0 0 o.1 0. 0 0.rad 0.3 0.2 0.75 and * = 9 0 * and various random error levels.75 0.3 0.4 Figure 5.
5 0 °.2 0.5 "(1O0°'P " /8*.rad Figure 5.2 .35 0..0..11 Allowable nonrandom error levels for y = 0.150.75 and = 1000 and various random error levels.1 0.25 .0 0 0.3 0.o. .3 0.4 o0.1 0.3d 0.3 0.1 45 0.1 0.d*)=0.2 0.4 /3.75 0 .0.4 0. 2 J 0. I O. 4 0.91 0.4 0.2 0.3 0.
5 10.d' )=0.4 .35 0.75 and = 1100 and various random error levels.45 0.1 0.1 0.3 I 0."1 I 0.4 0.92 0.20.2 0.3 0.rad Figure 5.4 0.0 0 0.1 0.G* .4 0.5 '. .4 0.12 Allowable nonrandom error levels for y = 0.2 0.3 0.75 o'a=0O.' '110o c y' ./0*.2 0.3 d* 0. I 0.2 I 0.
3 0.rad 0.2 0.1 0.1 0.3 0.13 Allowable nonrandom error levels for y = 0.3 0.5 ) =0.2  040.1 0.93 0.4 0.2 0.25 0.5 0.4 Figure 5.4 0.7 0.1 ' 0.01 0 0.5 Y(120 0 .2 /9.4 0. .2 0.5 04 0.1 0.75 and c = 120' and various random error levels.3 0.3 0. 0. o"*.3dm 120.
75 or 0.94 and horizontal intercepts of the curves represent the allowable level of one nonrandom error (d the other (6 or d ) is zero.14.95: : 0.the nonrandom error tolerance curves are not very sensitive to rim angle and the curves almost coincide . if d 0.e. Similarly. the changes in allowable level of nonrandom errors angle * (error tolerances) with respect to changes in rim are * shown. or when The curves themselves (solid and (when they lines) show the allowable levels of d both are nonzero).415 .e.2. * : 0. for 4 = 900 and y = 0.0 .15 and 5.0. d and y = 0.52 : o. (see Figure 5.5) < 0.0. As shown in Figure 5.0.1 rad.0.95). the allowable level of receiver location and reflector misalignment errors when they exist simultaneously.0 .14. then 8 < or for 0 = 0. i. the allowable levels of nonrandom errors (i.4 rad (see Figure 5.1).22 .16.. for a given rim angle.0.35 0.26 rad when a : 0. For example.1 rad.15.0 . say S= 90 '.13 for a specified (or required) minimum normalincidence factor (0.4 through 5.1 rad.0 . 5.75: when a d : 0.10) In Figures 5. for a given level of random errors. error toler ances d * and a * ) depend on the random error parameter O* and the rim angle 4. for small values of a (a = 0.0. As shown in Figures 5.
2 0.3d* y(.0 0 0.4 =o..2 0.1 So 100 0.14 Effect of rim angle on the allowable nonrandom error levels for a* 0.5 1000 . 14S .3 0.95 0.1 rad.1 0.1 0.3 0.d*) o.d*)75 05 0.1 0.95 0.1 0.O.4 0.2 0. 0.1.2 1 0.*.4 Srad Figure 5.3 0.1200 (. = .4 0.
3 * d" 0.' 100 1100 r ( .d*)=..3  0.2.2.1 0..3 0.1 0.d*) 0.1 0.4 Is*rad Figure 5.2 rad.O.G*.75 .4 9(f 0.96 0.4 12 Oe j=0.0.95 0.2 /1000 (4.jG*.2 0.2 /80* 0.3 I 0.1 0.00 0 0.4 0.5 0.5 .2 0.15 Effect of rim angle on the allowable nonrandom error levels for a* = 0. .2 12o00' 0.01 80 900 0.o.
2 1000 900 0.4 /* .16 Effect of rim angle on the allowable nonrandom error levels for a= and y = 0.4 0.4 0.2 0.0 1 0 0.35 rad .2 0.75 0.3 *.1 I 0.2 0.4 0.1 80o 0.3 0.3 0.0.1 0.5 C 0.d* )r 0.rad Figure 5.97 0.5 y (4S.35.3 d* 1200 1100 0.1 0.3 I 0.3 0.75 0.
98
(this is particularly true for y = 0.95 curves).
*
With
increasing a
values, the error tolerance curves start to
spread out and as a result the error tolerances start to be sensitive to the rim angles. For example, the changes in
allowable nonrandom error levels with repect to changes in rim angle are very evident in Figures 5.15 (y = 0.95,
* ,
a
=
0.2) and 5.16 (y = 0.75, a
=
0.35).
Therefore, one
can conclude that the range of optimal rim angles (which was found to be between 800 1200 for zero nonrandom errors)
would be different in the presence of both random and non random errors. In the following section, the sensitivity
of allowable level of nonrandom errors to changes in rim
angle is considered, and the optimal range of rim angles in
the presence of both random and nonrandom errors is found.
5.4 Sensitivity of Error Tolerances to Changes in
Rim Angle Optimal Rim Angles
*
In Figures 5.17, 5.18 and 5.19, sensitivity of al
*
lowable nonrandom error levels (d a
and
) is
shown for
given random error level (a*) and for specified (or
required) minimum intercept factors (y = 0.95 and y = 0.75). The locus of optimal rim angles for y > 0.75 are in the
range of:
800 
1200 for 950 
1200 for 1050 
1200 for
a a a
=
0.2 rad.,
0.3 rad.,
0.4 rad.,
=
99
80
90
100
110 0.
120
=Y 0.75
0.5
0.5
......
locus of optimal rim angles0.
0.3.0.
3
00.
0.~~
3
03
0.
.15
0.2

0.2
0.0
80 90
/,
0.1 10O0 110 120
Rim Angle,0, deg.
Figure 5.17 Allowable nonrandom error level d as a
function of rim angle and a* for a* = 0.2
and for y = 0.95 and 0.75.
100
80
0.5
90
1I
100
110
I
120
0.75 0.5
locus of optimal rim angles Y(6, 0.3,R3, d*)
40 .0.
0.3.0
0.2
0.2
0.1 •
0.1
0.0
80 90 100 110 Rim Angle,0, deg.
Figure 5.18 Allowable nonradom error level d
and Y = 0.75.
J0.0
120
as a
function of rim angle and a* for a* = 0.3
75 for a 0... 0.3 / =0.5 90 II 100 Y (p.1 .19 . '.B o" = 0.0 80 90 100 Rim Angle.2 0. as a = 120 Allowable nonrandom error level d function of rim angle and 8 and y = 0..1 0.4 .0 0.05 "".2 0.5 0.4 0.0 110 deg.101 80 0. 0.. 1 4 110 I 120 = d) 0..4 . 0.4. locus of optimal rim angles 0.75 0.3 0. Figure 5..0....
.2 locu. ocus of optimal l rim angles Y.102 80 90 100 110 120 0.95 0.20 . .1 0.3 0.. . 110 120 Figure 5.0...2 0.4 0. deg.2 0..ooptimal 0.75.0 1 80 1... 1 0.. 0.3 .1 0.. 0.45 .3.75 '= 0 Y"= 0..20 Allowable random error level d* as funtion a * * of rim angle and a 0.5 5 0. for 6 = .1 ..40 0.0 and y = 0.5 20.95 and 0./I 90 100 Rim Angle..
103 80 0.2 "0. o 0. .0 .2 0.0 0.7 O01 /0.3 y_ ...2 15 /.0 and Y = 0. 0 ..5 0.95 and 0.21 Allowable nonrandom error level 6 as a function of rim angle and a* for d* = 0.0 locus of optimal rim angles d= d 0.75.1 80 90 100 110 120 Rim Angle. deg. Figure 5.5  90 I 100 110 I 120 0.
. it is shown that the . 5. optimal rim angles are shown for zero angular nonrandom errors (i.20. found to be sensitive to both random and nonrandom errors. d 0. one can  rim angles are in the range of 1000 1200. The optimum rim angle based on both random and nonrandom errors was found to be in the range of 105'  120'. These are in the range of 80'.1200 for an intercept factor greater than 0.5 Summary In this chapter the results of the optical analysis obtained by using the comprehensive optical model and universal error para meters are presented. optical performance of the trough was.104 In Figure 5. It was concluded that the optical model developed in Chapter Three is a comprehensive model which can be used in a comp rehensive design method.21 optimal rim angles for zero mislocation non random errors (i. however. conclude that the optimum rim angle based on random and non random errors together is in the range of 105'120'.75.e. optimal As a result.e. These results were compared with those of the previous investigators and they were found to compare favorably. In this case. The results of the optical analysis with random errors alone were presented first.. The optimum rim angle based on random errors The alone was found to be braad (in the range of 800 to 1200).0).0) are shown. In Figure 5. 0. Furthermore.
(The concept of error tolerances is particularly useful in design of PTC's in developing countries. .) The use of the developed optical model will be presented in the second technical report.105 optical model developed here will enable a designer to incorpo rate error tolerances into the trough design at the preliminary stage of the design process. The following chapter presents models for the thermal analysis of PTC's.
depen ding on the temperature requirements of the application. Finally. In the process. however. a method for calculat per m 2 of ing a single heat loss parameter.1 Introduction As indicated earlier. For example. first.Chapter Six THERMAL ANALYSIS OF PARABOLIC rROUGH COLLECTORS 6. the absorbing surface of the receiver will be heated. the temperature difference between the ab sorbing surface and the surroundings will cause some of the collected energy to be transferred back to the surroundings (i.e. qL' in W receiver surface area to characterize the thermal behavior 106 . lost). and its temperature wili become considerably higher than that of the surroundings. Subsequently.. Proper quantification oT the heat loss from the receiver is important for predicting the perfoemance and hence designing PTC's. operating temperatures as high as 600OF can be attained at the absorbing surface of the receiver during operation. the primary function of the receiver subsystem of the PTC's isto absorb and transfer the concentrated energy to the fluid flowing through it. description of the problem and assumptions made for the analysis are presented. Then. models and empirical equations are presented for calculating the heat transfer coefficients between different components of the receiver subsystem. In this chapter.
a steadystate thermal analysis of the receiver will suffice for the design studies.ui and Assumptions The crosssection of the receiver subsystem is shown in Figure 6. working fluid will be heated as it travels through the fluid duct. a twodimensional steady The exchange wich the working fluid is shown. Heat exchange between the glass jacket and the surroundings. Tabs = Tabs (x). Heat exchange between the absorber tube and the glass jacket (glazing).e. thickness and conductivity of the absorber tube and the absorber tube temperature.. m. These are: 1.e. the value of the outlet tempera ture. three different heat exchanges the components of the receiver.. Therefore. i. Ta.1.and collector para for given Tfin. length of the collector module (L). will be a function of the inlet temperature (Tfin). 3. state energy In Figure 6. one can solve for . Heat transfer from the absorber tube to the working fluid.). etc. 6. exist among As shown. working fluid properties. Since the collector will be optimized based on either instan taneous or allday average efficiency. L.2 . flow rate (ri.). For a known heat flux along the absorber tube axis and for given inlet meters (i. Tout. and ambient conditions'.2 Desctiption of the Prob. which will be a function of x.107 of the receiver is presented. 2.
Absorber Tube (Tabs) Glass Jacket (Tgass) _ _ __ Annular Space Tf n .2 Working Fluid Twodimensional energy exchange to the working fluid [35].108 Turbulence Generator Absorber Tube With Selective Coating Annular Space Working Fluid Glass Jacket Figure 6. = 44444k X L 4 I Tf Turbulence Generator Figure 6. .1 Cross section of receiver subsystem.
flow rate. it will allow for the use of a onedimen . this is a very cumbersome method which requires numereous iterations and more importantly. further approximations are can be related to this average absorber tube temperature. more importantly.Ta) dx (6. Since at the preliminary stage of design only an average value of the heat loss per unit length of receiver will suffice. Then the total heat loss from the collector module can be calculated by using: L f 0 Uloss(x)(T glass(x) q0L = . However. Tabs(X) and Tglass (x) by setting up three energy balance equations [35]. this will eliminate a the need to specify the fluid inlet and outlet temperature. it requires a prior knowledge of the col lector parameters which are not known in the present case. fluid properties and collector parameters and.109 Tfout. The heat loss Therefore. it will be appropriate to assume a known average absorber tube temperature (Tabs = constant). and the heat loss coefficients for different modes of heat trans fer can be determined as a function of the average absorber tube temperature [361. (The objective of the present study is to optimize the col lector parameters!) needed.1) where Uloss(x) is the heat transfer coefficient for combined convection (wind induced) and radiation heat losses from the outer surface of the glass jacket. As result.
Therefore. In this model. The surfaces are gray. 6. 3. This is the model used in the literature most often (e. [35. 5. partially evacuated.. The surroundings acts as a blackbody at a reduced temperature Tk (Tnk (T 6)K). Steadystate conditions hold. heat balance equations can be setup at the inner and outer surface of the glazing as . The absorber tube and the glass jacket are con centric. the development of the onedi mensional heatloss model (using the above assunptions) is presented.3 OneDimensional HeatLoss Model The onedimensional receiver heatloss model is shown in Figure 6. diffuse emitters and ab sorbers in thermal radiative energy exchange. The problem is one dimensional in the radial direction.g.110 sional heat transfer model (radial).3. The annulus gas properties are functions of the temperature and pressure only. 7. The properties of air (surroundings). 4. 2. glass and selective coating are computed as a function of their temperature. 6. In the following section. the fol lowing assumptions can be made to facilitate the analysis of the thermal behavior of the receiver: 1. The annulus is assumed to be either nonevacuated (at atmospheric pressure). 8. or back filled with heavy gases. 36]) for the calculation of the heatloss from the receiver.
3 Onedimensional receiver heatloss model. Q2 = Condctive/convective heatloss rate from absorber tube to glazing. . Ta D g~o D D g. of heat input into glazing due to solar absorptance of glass. Conductive heatloss rate through glazing. Figure 6. 43 = = Rate Q = Convective heatloss rate from glazing to surroundings.o Tabs' abs Tg'0 031 ()3 = Radiative heatloss rate from tube to glazing.i Glass Jacket (Glazing) Cg. Qs = Radiative heatloss rate from glazing to surroundin .Absorber Tube Dg T g.s.i Pa .
only thermal conduction heat loss will exist in the annular space. Q: The radiation heatloss rate across the annulus can be calculated by the usual radiation exchange equation between two concentric cylinders (e. QLoss is then equal to: QLoss = Q5 + Q6 GLoss (6. Q2 : Natural convection heatloss in the annular space will be negligible as long as the Rayleigh number [38]. See next page for the definition.. . is less than 1000 Therefore.5) L 1 abs + D g.i 1 g.3) are modeled.i 1 Conductive/Convective heatloss rate from absorber tube to glazing. Modeling of the heatloss rates is described in detail below per unit length of receiver: Radiative heatloss rate from absorber tube to glazing.112 follows: Inner surface: QI + Q2 Q3 = (6.4) can be obtained by iterative solution of the Equations (6.2) and (6.3). see.3) The heatloss rate from the absorber tube.g. for Rayleigh numbers less titan 1000. after all the heatloss rates in Equations (6.2) and (6.2) Outer surface: Q3 + Q4 = Q5 + Q6 (6. [371): QaDo (Tbs  gi (6.
the heatloss rate can be calculated from Eq. 40]). The cl:. therefore. kef.7) by using the conductivity of the annular gas at reduced pressure (e.113 and the heatloss can be found from: Q2 L L 2 Tr k gas kgas DgjD ) ( (Tabs  Tg) < 1000 (6. Q 2 = 0.5 (T +T .(6. T m = 0. correlation for kef is given here (see Eq.ssical experimental results of Kraussold have been correlated and used by many investigators..1585 kgas Ra Na> 1000 Ra26 00 (6. Heat transfer Dg i NRa= Rayleigh numberP g D p(T gi abs . This Q2 L = 2 ksef Dg i (T)T(6. experimental results reported in literature (e. For evacuated annulus k e * 0. which can be obtained from [39.8)).0.6) ln( where kgas is thermal conductivity at standard conditions..7) a abs g.g.(6. evaluated at mean annulus temperature.g.8) 68 For partially evacuated annulus.) " For Rayleigh number*greater than 1000. the combined ccn duct ive and convective heatloss rate can be evaluated from an effective conduction coefficient [38]. see [41]).i D) where kef =o.
it can be assumed constant. CT T ) wind .o a ( 0 (6.9) in Dg'° Dg.: Q 4 L where glass I b P W p = reflectance of the reflective surface. Ib.i Rate of heat input into glazing.o (6. However.10) where Nuwind. Q 4 This will be a function of the beam insolation. Q 5 This will be a function of the wind velocity: = L k air Nu. = wind velocity induced Nusselt number = kair . Q3 This is given by the following equation: Q3 L 2 T kglass (T g'i T g. Conductive heatloss rate through glazing. therefore. Convective heatloss rate from glazing to surroundings. W = aperture width of the reflector. the sensitivity of the overall heatloss coefficient to Q4 is found to be small [36].114 in the annular space will be discussed further in a later section.
(cross Note that all prop erties are evaluated at T m = 0.. see Figure 6.5 (T a + T go). SVWind (a) CrossFlow Figure 6. Q 6 This can be calculated from the following equation.4b).4 (b) ParallelFlow Wind induced heatloss from receiver. at a reduced temperature (Tsink = (T .15 The heat loss coefficient will be a function of Reynolds number and its value can be obtained using Hilpert's for mulation for forced convection over cylinders [42] flow over cylinders. the heatloss rate Q 6 per unit receiver length is: There .4a). At present.e.6) K). flow along the axis of the tube. see Figure 6. V Wind 0. there are no existing studies for calculating the heat loss rate for cylinders in parallelflow (i. Note that the surroundings are assumed to act as a blackbody 0 Therefore.3. Radiative heatloss rate from glazing to surroundings. fore only crossflow case can be analyzed.
that is based on absorber tube surface area. 6. Pa' Ta. radiative. Initial estimates of T 9o and Tg.T ) or L (6. Egi. Qloss/ L. Receiver heatloss is related to U L by the following equation: QLoss (watts) = U L 7 D L (Tabs . (6.T a [W/mabsoC] ) 7DL and. UL. Equations (6. D.o 7t Dg o (T.i.4 Calculation of the HeatLoss Parameter qL Given the receiver type (nonevacuated.i will be made and then corrected iteratively to obtain the unknown tempe ratures.  Ts4nk) (6. therefore.11) where a is the StephanBoltzmann constant. and fluid and gas properties will be varied also at each iteration. the heat loss parameter qL (in W per m 2 of . kglass. Temperaturedependent convective.4). Eabs (T). Eg. Receiver heatloss can then be expressed as a heatloss coefficient. and Tab s .116 Q 6 g. Vwind.13) Tabs .11) can be used to calculate the heatloss from the absorber tube per unit length of receiver.10) and (6.o. backfilled with heavy gases). Once T9o is found. evacuated. Equations (6. Dg.o.12) QLoss (6.3) will be solved simultaneously by iteration.2) and (6. Dg.
91 to 3. 0. 14  Absorber Diameter Range 1.117 TABLE 6.18 cm 12 10.90 0.15(100 0 C).i Eg0 0.91c9 10 Reference Trough 8 Receiver 3.25(300 0 C) at cabs (Blackchrome) Blackchrome emittance assumed linear between and beyond these limits.18 cm SEvacuated _j1.1 REFERENCE TROUGH RECEIVER DESIGN PARAMETERS Parameter Value Eg.91 1 0 _ _ I I cm I I I II Receiver 0 50 100 150 200 250 300 350 Averaqe Absorber Tube Temperature (0C) Figure 6.95 0. .5 Heatloss coefficient UL as a function of average absorber tube temperature and absorber tube dia meter (from Reference 36).
In this example.118 receiver surface area) can be written as: qL = TrD s UL ( TZ [36] . ambient temperature and wind ve locity using the equations given in the preceding section. Curves similar to the ones shown in Figure 6.14) For example.5 Annulus Gap Sizing In typical high temperature absorber tube designs. The reference receiver parameters are given in Table 6. Figure 6. Gee et. the annular gap is assumed to be sized such as to keep Rayleigh number in the range 0 . al. and can amount to approximately 6% of the total rate at which energy is absorbed by the collector [38].5 will be used in the second technical report to solve example problems. A yearly average ambient temperature of 10'C is assumed.5 can be obtained for other re ceiver design parameters.5 mph) over the receiver is assumed as representative of average wind conditions.5) using such a procedure. Therefore elimination or .1. A wind velocity of 2m/s (4. 6. the rate of energy loss by combined thermal conduction and natural convection in the annulus is of the same order of magnitude as that due to thermal radiation.1000. have calculated the heat loss coefficient.T ) (6. UL as a function of average absorber tube tempe rature and absorber tube diameter for a "reference" receiver and an evacuated receiver (Figure 6.
i. 44. They are: * Backfilling the annulus with a heavy gas (gas with low thermal conductivity). Figure In sizing of the annulus gap is shown for [41].119 reduction of conduction and natural convection losses can significantly improve the performance of a collector field. As mentioned earlier three different techniques can be used for reducing or eliminating these losses. maintaining a vacuum in the annulus was found to be very difficult. 45] have investigated these heat reduction techniques extensively as part of Sandia's PTC development program. This corresponds to a 45 percent improvement in overall collector efficiency [43].. Some of their conclusions are: * Overall collector efficiency could be improved by 11 to 12 percent if annulus pressures can be maintained below 10 2 Pa [35]. an absorber temperature of 3150C In Figure 6.7 sizing of the annulus gap for an absorber (Reference trough re As shown. Rayleigh number should be less than 1000. * The gap size should be made as large as possible while suppressing convection in the annulus. * Evacuation. However. [43.) cuated receiver. * Heavy gas utilization in the annular space can reduce receiver heatloss by 50 percent. * Oversizing the annular space (to minimize conduction losses). for an eva temperature of 2000C is shown [36].1. Ratzel et. annulus gap sizing is not thermally significant because no conduction or convection occurs in the annulus. . ceiver data is given in Table 6. al.e.
90 1.60 0.2 5IIII 245" 230 II Absorber Temperature = 315 0 C U) 230 Carbon Dioxide 215 0 CU 200 Air • o 170V 1 Onset of Convection Argon 155 0. 41 absorber temperature of 315'C (nonevacuated receiver). .45 0.20 1.05 1.75 0.6 Annulus gap sizing or alternate gases using a 2.54 cm absorber tube for an Source: Ref.35 Annulus Gap (Cm) Figure 6.30 I 0.
It was shown that the presented thermal analysis could be used to size the annulus gap size.0 Absorber to Glass Gap Size (cm) Figure 6.5 3. . It was shown that this model could be used to calcu late a heatloss parameter qL in W per m 2 of receiver surface area to characterize the thermal behavior of the receiver. examples in the This is demonstrated by illustrative second technical report.121 14  Absorber Temperature = 200 C Reference Trough Receiver Xenon BackFilled Receiver Receiver I P 00% Go\ 121 108 r 6  C4JC _Evacuated D) 2 4 C 0 iI ! 0.5 2.0 2. temperature 6.7 Annulus gap sizing for an absorber of 200'C (from Reference 36). The method developed in this chapter can be used in a comprehensive design and optimization method.6 Summarv In this chapter.0 1.5 1. a onedimensional heat transfer model for the thermal analysis of the receiver subsystem was pre sented.
The use of the results of the thermal analyses available in the literature in a comprehensive design method was discussed too.Chapter Seven CLOSURE The need for comprehensive models for optical and thermal ana lysis of PTC's to be used in comprehensive design studies was established in this report. groups as random and nonrandom. In this report. and absorber diameter D) to yield error parameters universal to all collector geometries.). It was shown that available PTC optical they had restrictive assumptions in a comprehensive design method models were not comprehensive. manufac ture. 122 it was shown that the . These parameters were called 'universal error parameters'. The results of the developed optical model were presented using the universal error parameters. Furthermore. the development of a comp rehensive optical model for PTC's was presented. C. error parameters were combined with geometri( parameters (concentration ratio. assembly. A thermal model suitable for design and optimization purposes was also presented. It was concluded that the optical and thermal models presented in this report are comprehensive models which can be used in a comprehensive design method. which limited their usefulness which could lie used for design of PTC's in developing country design environments. Errors were divided into two One random and two nonrandom error parameters were used co characterize different kinds of errors. etc. materials. In the presented optical model particular emphasis was placed on the modeling of different kinds of errors (operational. Subsequently.
(No such data was found to exist in the literature.123 optical model will enable a designer to incorporate error toleran ces into the trough design at preliminary stage of the design process.) *Such a study will help validate the modeling of the nonrandom errors presented in this report. An experimental study to investigate the effect of nonrandom (systematic) errors on the performance of the trough is highly recommended. A study to investigate the wind induced parallelflow heat loss from receiver (see Figure 6. The concept of error tolerances is particularly useful when designing PTC's for developing countries.4b) is also recommended. .
1930. v. 15. 10. University of Houston . G.. N. J. G. May 1975..1 (1955). N5C. July 1975 124 . 12. v. Albuquerque. Mechanical Engineering. Tabor.M. W.. Sandia Laboratories. Fester..University Park. D. August 1974. Albuquerque. of Solar Energy Engineering. 104:209.. "Performance of a Focusing Cylindrical Parabolic Solar Energy Collector: Analysis and Computer Program" SLA740031.. Doctoral Dissertation. N.. "Solar One". SAND. 24. A. Vol.... F. J. G. N. "Theoretical Performance of Cylindrical Parabolic Solar Concentra tors". 6.. 219244.. 2. H. "Solar Energy Collector Design". N. Kreith. Treadwell.M. Sandia Laboratories. 1962.REFERENCES 1. W. 3. " A Comprehensive Method for Computer Aided Design of Practical Energy Systems". Edenburn. Hauil M. 5..755333. Sandia Laboratories. Solar Collector". Transactions ASME Series A 84. 102:174. "Test Results from a Parabolic Cylindrical.. No. Albu querque. H.. No. Hassan. pp. 7. W. KamalEldin. 4. of Solar Energy Engineering.. W. "Can Industry Afford Solar Energy?". Duffie. 8. 1980. H. Treadwell.. R. "Energy Balance on a Parabolic Cylinder Solar Collector" Journal of Engineering Power. Bezdek. SAND 740124. G. W. 1982. December 1983. Albuquerque. March 1983 3. M. "Solar Energy for Developing Countries". W. "Selection of Parabolic Solar Collector Field Arrays" SAND740376. Banas. R. N.. 0. J. Treadwell. Sandia Laboratories. M.M. Department of Mechanical Engineering. F. F. July 1974 9.M. G. Gi*ven. Solar Energy. McCulloch. A. and ElRefaie. Rusk. September 1973. S. L f. Bulletin of the Research Council of Isra:_i. J.. McCulloch. Kreith. Sandia Laboratories. Kreith.. "Technology Assessment: LineFocus Concentrators" SAND 792221. "Design Analysis cf Asymmetric Solar Receivers". F. F.M. 11.
pp 1621.M. Vittitoe.. "Receiver Assembly Design Studies for 2m 90* ParabolicCylindrical Solar Collectors.M. 22. Poon. N. and Carley. Ratzel. L. 1977. D... Aug. 631. Treadwell. "Laser Ray Trace and BiDirectional Reflectometry Measurements of Various Solar Concentrators.. Grether.." Transaction of ASME. W. Sandia Laboratories. Vol. N. Evans. Golden. H. 1980. W. 21. pp.. A. C. C. 14..." SAND 760082." SERI/TR 34092 Solar Energy Research Institute." SAND 791026. "Simplified Calculational Procedure for Determining the Amount of Intercepted Sunlight in an Imaging Solar Concentrator. "The Helios Model for the Optical Behavior of Reflecting Solar Concentra tors. Pettit. Gaul. and Biggs.B.. F. "An Analysis of the Influence of Geography and Weather on Parabolic Trough Solar Collector Design.M.. Wen. 18. "On the Performance of Cylindrical Para bolic Solar Concentrators with Flat Absorbers. R. N. Albuquerque. 19.640. A. C. L. 1979." Transactions of ASME. Journal of Solar Energy Engineering. 102. Albuquerque.. Rabl. "IncidenceAngle Modifier and Average Optical Efficiency of Parabolic Trough Collec tors. L. A. Vol. 17. Vol. G... J. and Rabl. Huang. A. Treadwell..M.. November 1980. D. "Transactions of ASME. and Reed. or see [27] PP. 123. and Biggs. N. pp 379385." Lawrance Berkeley Lab. 1979. R. J.. 102. R. 105. Grandjean. W. 23. Pettit. Sandia Laboratories. "Opti cal Analysis and Optimization of Line Focus Solar Collectors.. Georgia Tech.. Gaul. pp. 1976. 1978. CO.. 19." Solar Energy. Sandia Laboratories.. 16. and Butler. Mar. Albuquerque. L... H. Sandia Laboratories. N. and Vittitoe.. G.125 13. Albuquerque. K.. Biggs.. A. B. N. and Hunt. B. 9. F." Solar Concentrating Collectors. Feb. F. 20. P. "Design Considerations for Parabolic Cylindrical Solar Collectors. 305315. of Solar Energy Engineering Vol. 1983. 1980. Bendt. "Description of the LBL Reduced Data Base and Standard Profiles." SAND 760347. 15. of Solar Energy Engi neering." SAND 79 2032. P. N. "Compara tive Study of Sclar Optics for Parabolodial Concentra tors. . September 1979. 1977..
. Gaul. 35. Vol.. "Optical Evaluation Techniques for Reflecting Solar Concentrators.. B. 33.M. S. 1977... Ratzel. . and Workhoven.. V. 114. Sandia Laboratories Albu querque. Thermal Radiation Heat Transfer." in preparation (to be published in 1985). Optic Applied to Solar Energy Conversion.126 24. Biggs. of Solar Energy Engineering. 102. W. J. A..M. Academic Press.. and Grandjean. R. "Determining the Optical Quality of Focusing Collectors without Laser Ray Tracing. 407417. R. S. Treadwell. Pettit. No. 2. 34. J. Rabl. H. pp. B. Dudley. Vol." SAND 810159. (1976).. 109113. 1983. G. Vol.M..." Solar Energy. 5. "Summary Report: Concentrating Solar Collector Test Results Collector Module Test Facility. "Characterization of the Reflected Beam Profile of Solar Mirror Materials.." Albuquerque. "Optimization of Parabolic Trough Solar Collectors. R. and Moustafa. A. Hl. R." Transactions of ASME. C.M. M. "Comprehensive Error Analysis and Derivation of Universal Error Parameters for Paribolic Troughs. and Petit. May 1980. N. No. 28. Medium and High Temperature Solar Proc esses. Bend. 1972.. 30. D. Vol. A. "Geometric Effects on the Performance of Trough Col lectors." unpublished notes provided by G. Siegel. Kreider. R. McGrawHill. 31. 27. No. 1982. Jarrar. Sandia Labo ratories.. GUven. A. 30..128133. F. R. I. "Evaluation of the Evacuated Solar Annular Receivers Used at the Midtemperature Solar Systems Test Facility (MSSTF)". april 1981. EDEP: "A Computer Program for Modeling the Parabolic Trough Solar Concentrator. N. L." Solar Energy. H. W. Jeter. and Gaul. Sandia Laboratories.. 25.. N.. 6. R. 29. W. Treadwell. Albuquerque. and Bannerot. pp. and Howell. SAND 780983." Solar Energy. P. 26. 19. P.M. J. B. July 1979. 1979. and Rabl. Butler.. N. E. Albuquerqe. 29. "Systematic Rotation and Receiver Location Error Effect on Parabolic Trough Annual Performance. 32. Bendt. M. pp. Sandia Labo ratories. 1977. N. F.B." Society of PhotoOptical Instrumentation Engineers (SPIE) Vol. 1978.
pp. Rabl. Solar Energy Research Institute.. "Tech niques for Reducing Thermal Conduction and Natural Con vection Heat Losses in Annular Receiver Geometries".M. 5.127 36. Geb. Ratzel.. . Mirkovich. Heat Mass Transfer. Albuquerque. a. 1978. R. A. J.. 1934. pp. H.M.. March 1980.. 4. 1979. J. Ratzel. W. Kearney.. and Goldstein. Gaul. Vol.K. R. ASME paper No. Hilpert.. "Wbrmeabgabe von zylindrischen FlUssig keitsschichten bei natirlicher Konvektion".. E. and Gartling. D. February 1979. Forsch. pp. Vol. Ocober 1980. 45. H.. vol. 4.. p. Kreider. Ratzel.. D. SERI/TR632439.. "Correlating Equa tions for Natural Convection Heat Transfer Between Horizontal Circular Cylinders". J. 1980. Kraussold. Keuhn. 39. Ratzel. McGraw Hill. C. 37. A.. Hickox. Dt. 41. vol. Golden. R.. E. Principles of Solar Engi neering. "Energy Loss by Thermal Conduction and Natural Convection in Annular Solar Receivers".. Sandia Laboratories... C. 42. 79WA/Sol18. C. 101. In genieur wes. Simpson. C. "Utilization of Heavy Fill Gases in Annular Solar Receiver Geometries for Heat Loss Re duction". Gee. 11271134.d. K.108113. No. Kreith. Ratzel. A.215. SAND781769. 38. d. Hickox. Edited by Vladimir V. Sandia Labora tories. D.. 1978. "Long Term Average Benefits of Parabolic Trough Improvements"... N.. of Heat Transfer.. Albequerque. A. N. Thermal Conductivity.. C.. J. "Warmeabgabe von geheizten Drahten und Rohren im Luftstrom". Forsch Hft. "Heat Loss Reduction Techni ques for Annular ScAJar Receiver Designs".. 186191. 40. C.. SAND 791010. C. 19.E. Ing. 43. and Gartling. A. Plenum Publishing Company. 44.Y. C. N. F. Ver. F. C. Int. T. 1933. "Annular Solar Receiver Thermal Characteristics". A. 1976. Co. H. Sisson. New York.