SECTION 4 - UAB · 2018-07-24 · Ketan Solanki, Mohamed Nizamuddin Shaik, Vijaya Krishna Teja Bangi Department of Mechanical Engineering Lamar University Beaumont, TX, USA Dr. Ramesh

UAB School of Engineering – Mechanical Engineering - ECTC 2015 Proceedings – Vol. 14 Page 107

SECTION 4

Heat transfer,

ENERGY GENERATION

& CONVERSION



Proceedings of the Fourteenth Annual Early Career Technical Conference The University of Alabama, Birmingham ECTC 2015

November 7, 2015 - Birmingham, Alabama USA

DESIGN AND STUDY OF SOLAR CONCENTRATOR WITH FLAT HEXAGONAL PETALS

Ketan Solanki, Mohamed Nizamuddin Shaik, Vijaya Krishna Teja Bangi Department of Mechanical Engineering

Lamar University Beaumont, TX, USA

Dr. Ramesh K. Guduru

Department of Mechanical Engineering Lamar University

Beaumont, TX, USA

Dr. Kendrick T. Aung

Department of Mechanical Engineering Lamar University

Beaumont, TX, USA

ABSTRACT

The rapidly increasing demand for energy along with

depletion of global fossil fuel reserves makes renewable energy

a very attractive long term solution. Among various renewable

technologies, the harnessing of solar energy for generation of

electricity and heating applications is certainly one of the

foremost and most easily viable solutions. There are different

approaches in practice for collection as well as conversion of

solar energy into many useful applications. Solar flat collectors,

solar troughs, and tube collectors are conventionally used for

several engineering and household applications. Along these

lines, we have conducted a new feasibility study on use of

Aluminum (Al) kitchen foil and hexagonal petals for a solar

concentrator with an aim to reduce the costs and complexity of

solar dish fabrication. Here, we present a design and

experimental study of a solar dish concentrator built using

Aluminum (Al) foil wrapped around flat hexagonal cardboard

petals as a medium of reflection. The solar concentrator was

designed based on dish geometry for high collection and

concentration efficiency in addition to cost effectiveness.

However the arrangement of flat petals into a dish geometry

was dictated by the angle () between the petals as well as the

petal size (a) while limiting the amount of area of reflection on

the dish. We made several 3D designs using CATIA V5 with

variations in ‘’ values and petal sizes in order to determine a

compatible arrangement of the flat petals. Among various petal

sizes and ‘’ angles, we chose to build a prototype with petal

size of 25mm for an angle () of 30

in our experimental

investigation, using the CATIA designs in order to have more

area for reflection. This experimental setup yielded a maximum

temperature rise of 1200C under an average solar radiation of

1000 W/m2. This is the first work ever to report a temperature

rise of 1200C with use of aluminum kitchen foil and hexagonal

petals on a solar concentrator.

INTRODUCTION The most abundant energy source for the earth is the Sun.

It provides around 3.8 x 1020

MW of energy, which is

equivalent to a power of 63 MW/m2. However, due to reflection

and absorption by the atmosphere, only 174 Peta Watt is

reached [1]. If this energy is totally captured for an hour period,

it can easily supply the energy needed for the whole world for

more than a year.

Solar energy can be utilized in many ways, and currently it

is being used for water heating and generation of electricity in

the households. The worldwide demand for electricity has been

increasing by 5% every year [2], and therefore relying on

nonrenewable (e.g. fossil fuel) sources will eventually hamper

the human life with depletion of resources in the long term.

However, use of renewable energy sources with cost efficient

technologies will not only boost the effectiveness of solar

energy consumption but will also reduce greenhouse emissions

and help in sustainable growth.

Solar concentrators are certainly one of the most attractive

technologies in the household applications, with a good room

for improvement in efficiency of operation while increasing

their affordability, in contrast to the solar cells. Among the

different types of solar concentrators [3], the parabolic dish is

the most efficient system for solar energy concentration

because of its high collection and concentration effectiveness.

However, the area of a solar concentrator’s field is dictated by

its size and the amount of energy collected. Usually, dish type

solar concentrators are made using curved mirrors [4] for solar

cookers and to generate electricity [5]; curvy hexagonal petals

are used for reflector dishes in telecommunications [6]. But,

these are not affordable to the common man due to expensive

manufacturing costs associated with complex fabrication

approaches. Thus, there is certainly a huge requirement for the

development of inexpensive and efficient technologies that can

capture the solar energy for common applications with greater

affordability.

Keeping those above discussed goals in mind, here we

develop a solar concentrator based on dish geometry using flat


hexagonal petals. Initially, we chose to make CATIA models

with hexagonal petals of different sizes (a) and different angles

() between the petals as shown in Fig. 1. After analyzing the

petal arrangement and the area of reflection for a combination

of different petal sizes and the angles between them, a

prototype was developed and subsequently implemented for

experimental investigation. These studies are unique in terms of

using inexpensive materials, such as Al kitchen foil and flat

cardboard petals, to develop a solar dish concentrator. The

experimental observations yielded a high temperature rise of

120 0C.

Figure 1. Schematic for a solar concentrator dish showing dish diameter, petal size (a) and the angle between the

petals ()

METHODS The quantity of solar energy reaching the Earth's surface

averages about 1,000 W/m2 under clear skies, depending upon

weather conditions, location and orientation. Solar collectors

are typically used in residential and commercial buildings for

space/water heating.

In our approach, the following parameters were considered

for the design of the parabolic dish

1. Area of the dish.

2. Reflectivity of the material.

3. Hexagonal petal dimension (a).

4. Angle between the petals ().

5. Tolerance/gap between the petals.

The area of a dish is primarily controlled by the

dimensions of the petals, the angle between adjacent petals, and

the tolerance maintained between the petals. In order to obtain

optimum reflection of the solar radiation, the surface of the dish

is supposed to be as smooth as possible, and this can be

achieved using petals of smaller dimensions. Also, as the angle

() between the petals grows small, the dish area increases in

size, and vice versa.

The arrangement of petals in a dish is dictated by the

tolerance or gap maintained between them, and it will

eventually make the petals run in to each other, causing a force

fit or termination of the dish symmetry, depending on the petal

size. However, the gap between the petals can be varied

accordingly to achieve the best possible fit. A smaller gap leads

to early termination, while the larger gaps compromise the

effective surface area of reflection, depending on the

dimensions of the hexagonal petals. Thus a tolerance

maintained between the petals will play a key role in the total

reflective area of the dish.

3D Modeling We used CATIA V5 to design our solar dish models with

flat hexagonal petals. The strategy used in our designs was to

achieve a maximum reflective surface area. The flat hexagonal

petals were placed starting from the center of the dish, and 1

mm tolerance was chosen between the petals. As the petals

were arranged from the centre of the dish, the gap between

them gradually reduced. Finally, the stacking of petals failed

when there was no gap between them.

Experimental approach In a solar concentrator all the sunrays are reflected on to a

defined focal point to concentrate the solar energy. In our

experimental studies, the dish was fabricated with incoming

light rays parallel to the dish's axis to reflect onto an absorber

that was placed at the focal point. The dish was aligned with its

axis pointing toward the sun, allowing almost all of the

incoming radiation to be reflected onto the focal point.

Radiation losses in such collectors could occur due to scattering

of the light in to a wide range of angles, if all the petals are not

focused onto a single spot. Also, the type of reflective materials

used, such as aluminum foil, mirrors or any reflective painting,

will control the extent of light reflection; Al kitchen foil has

reflectivity around 85% [7].

RESULTS AND DISCUSSION

3D Design In our design process, two distinct design approaches were

followed. In the first approach, the angle between the petals ()

was maintained constant while their size (a) was varied; this

approach enabled us to fix petals of different sizes in a dish

with fixed dimension i.e., fixed diameter and angle ‘’. In the

second approach, the petal size (a) was fixed, and the angle ‘’

between the petals was varied while keeping the diameter of the

dish constant. This helped us to determine the maximum

possible reflective area of the dish for different ‘’ values.

Following the first approach, we created models with

different petal sizes (a) for a given angle () of 100. Table 1

shows the variation in reflective surface area on the dish as a

function of petal size.


Table 1. Variation of petal size and reflective surface area

for a fixed angle () of 100 between the adjacent petals

Sr. No

Size of Petal (mm)

Angle (Degree)

()

Number of stacking layers

Reflective Surface Area

(mm2)

1 30 10 1 16366

2 25 10 2 30856

3 20 10 2 19741

4 15 10 2 11115

5 10 10 4 9620

Figure 2 shows the stacking of petals with a size of 30 mm

for an angle () of 100 between the petals, which failed at the

second layer of stacking, causing them to run into each other, as

shown by the arrow

Figure 2. Arrangement of hexagonal petals in to a dish

with a petal size (a) of 30 mm and an angle () of 100

between adjacent petals

In a similar way, the petals arranged in a dish for a petal

size of 5 mm are shown in Fig. 3. However, this configuration

failed at the fifth layer of stacking. The gap of 1 mm between

the petals in the initial iteration enabled for the accommodation

of improper symmetry as the stacking continued. However, it

failed after the fifth layer of stacking, and thus proved the

important role of the gap between the petals in achieving a

maximum surface area, while facilitating increased numbers of

layers of petal stacking before they crash.

For every petal size, the total reflective area on the dish

varied, depending on the number of layers of stacking. Figure 4

shows the variation of total reflective area on the dish with

changing petal size. Based on these designs, and from the Fig.

4, it is clear that the highest reflective area was achieved for a

petal size of 25 mm.

In the second approach, we kept the size (a) of the

hexagonal petal constant, and varied the angle between adjacent

petals (θ) so that different sizes of concentrators with varying

diameters could be modeled. However, choosing an optimum

angle was very tricky. If the angle was larger, the focal point

would be close to the dish and thereby the reflected area would

decrease. The problem with this configuration was that the

absorber placed in the focal point could block a considerable

amount of radiation and would compromise the efficiency of

the design. On the other hand, a smaller angle would result in a

focal point that was farther from the dish, which also would

have an effect on the solar concentration of the system. This

will be discussed momentarily in regards to the blowing winds

as well as scattering of light into a wider angle range.



between the adjacent petals

Figure 4. Variation of reflective area on dish with different

petal sizes for a fixed angle () between the petals

Figures 5 and 6 show the configuration of 10 mm petals for

angles () of 200 and 3

0, respectively. In Fig. 5 the arrangement

of petals crashed in the second layer of stacking because of a

very high angle of 200 between the petals. Using the smaller

angle of 30 resulted in a maximum reflective surface area of

86060 mm2. However, this design could present challenges for

fabrication, as the petals must be manually fixed into the dish

geometry to focus on a single point, and their smaller size

presents problems with scattering light into a wider angle

range.

16366

30856

19741 15998

11115

9620 3965

0

10000

20000

30000

40000

0 10 20 30 40 Su

rfa

ce

Are

a (

mm

2)

Size of Petal (mm)

Size of Petal (cm) Vs Surface Area (mm2)


Table 2. Variation of the angle between the adjacent

petals () for a fixed petal size

Sr. No

Size of Petal (mm)

(a)

Angle (Degree)

()

Number of stacking layers

Reflective surface area

(mm2)

1 10 20 2 4940

2 10 18 2 4940

3 10 15 3 9620

4 10 12 3 9620

5 10 10 4 15860

6 10 5 7 33020

7 10 3 10 86060







Figure 7 shows the variation of total reflective surface area

for a 10 mm petal size on the dish for different angles between

the petals, which certainly proves that a smaller angle would

result in a higher reflective area.

Figure 7. Variation of dish area with different angles () between the petals for a fixed petal size (a= 10 mm).

While investigating for the optimum dimensions of the

hexagonal petals and the angles between them, it was observed

that smaller petal dimension with smaller angles resulted in

much smoother concentrator surface with more layers of petal

stacking. The design data from the Tables 1 and 2 prove this

notion. In practical terms, it is difficult to fabricate smaller

hexagonal petals. Also, a smaller angle between the petals

implies a larger dish area, and the focal point would be farther

away from the dish. In such cases, the solar energy

concentrated will be comparatively less because of scattering of

the reflected radiation and also because of the convective heat

losses from the absorber surface to the surrounding

environment. Therefore, we chose to further investigate our

designs with different petal sizes for an angle of 3º to achieve a

maximum reflecting area.

Table 3 shows the highest reflective area for a petal size of

25 mm with an angle of 30

between the petals. Based on this

design we conducted our experimental investigation.

Table 3. Variation of petal size for a fixed angle 30

between the adjacent petals ()

Sr. No

Size of Petal (mm)

Angle (Degree)

Number of

iterations achieved

Reflective Surface

Area (mm2)

1 30 3 3 86506

2 25 3 5 147784

3 20 3 5 94549

4 15 3 6 74295

5 10 3 10 86060

FABRICATION OF CARDBOARD DISH

Materials used- Cardboard of 2 mm thick, Al kitchen foil

and glue/adhesive tape.

Design criterion-From the CATIA designs, a maximum

reflective area was achieved when the petal size was 25 mm for

an angle of 30 between adjacent petals. Hence, we chose to

fabricate the dish of the same dimensions/geometry.

4940 4940

9620 9620

15860

33020

86060

0

20000

40000

60000

80000

100000

0 10 20 30 Su

rfa

ce

Are

a (

mm

2)

Angle (degree)

Angle (degree) Vs Surface Area (mm2)


Fabrication of dish- Hexagonal petals of 25 mm were cut,

and Al kitchen foil was glued on to these petals to prevent

wrinkles. The petal was then allowed to set for 2 hours before

being fixed on to the dish. The frame work for the dish was

made using cardboard sheets as shown in figure 8. In this

design, the focal point was set at 250 mm with a diameter of

200 mm.

Figure 8. Supporting base structure for the dish.

The base for the hexagonal petals was developed by

sticking duct tape to the supporting structure shown in Figure 8.

Figures 9 and 10 show the 3D design and construction of a

prototype, respectively, for a petal size of 25 mm and 30 angle.




Figure 10. Arrangement of hexagonal petals on a dish



TESTING OF SOLAR CONCENTRATOR The prototype concentrator was exposed to incoming sun

radiation on a sunny day with solar intensity around 1000

W/m2, and a maximum temperature of 120

0C was obtained.

See Figure 11. The temperature readings at the focal point were

taken at regular intervals of two minutes, and the experiment

was continued for half an hour. These experimental

investigations verify our modeled design in terms of stacking of

petals, and thereby support the design approach. The

experimental data yielded a reasonable temperature rise while

utilizing flat petals in to a dish geometry.

Figure 11. Variation of solar radiation and temperature at

focal point with respect to time (a=25 mm and =30).

EXPERIMENTAL OBSERVATIONS The goal of this project was to study the feasibility of

developing a flat hexagonal petal based dish for solar

concentrator applications, using inexpensive resources

available to us. Our study was done using Al ktichen foil as a

reflective medium. Preliminary experimental observations with

a prototype showed very promising results with possibility to

achieve temperatures up to 120 0C using mere Aluminum foil

reflectors (85%) [7], in contrast to the expensive and highly

reflective curved mirrors (95%) [8].

While testing our prototype, various un-controllable

problems were faced, such as blowing cold air and frequent

blockage of sunlight by clouds from time to time causing

fluctuations in the radiation collection and thereby resultant

temperatures. A mechanism for accurately tracking the position

of the sun has yet to be incorporated in to the system.

To avoid such uncontrollable problems, further research in

this direction can be done by use of glass dome over the solar

concentrator to avoid convective heat losses due to wind. A

mechanism for tracking the Sun can be developed and

incorporated into the system to ensure a constant input of solar

radiation with respect to the time.

CONCLUSIONS We conclude that the methodology and design approach

followed in the present work can be implemented to develop

inexpensive solar concentrators using regular Al foil and

50

70

90

110

130

1000

1010

1020

1030

1040

1050

0 10 20 30 Te

mp

rea

ture

(0C

)

Inte

nsity (

Wa

tt/m

2)

Time (min)

Intensity Of Radiation Tempreature


cardboard petals. These studies shed light on enhancing

renewable energy technologies and design opportunities for

large scale applications. However, there are some limitations of

this technology with direct exposure of the radiation collector

to the winds and also the pliability of the framework and petals,

materials, designs and fabrication methods. The results prove to

be promising to obtain high temperatures with very low cost of

manufacturing, making this proof of concept design successful.

REFERENCES [1] Kreith F, Kreider JF., 1978, “Principles of solar

engineering,” New York: McGraw-Hill.

[2] World watch Institute, 2009, State of the World – Into a

Warming World.

[3] Soteris A. Kalogirou, 2004, “Solar thermal collectors and

applications,” Progress in Energy and Combustion Science 30,

pp. 231–295.

[3] A. R. El Ouederni1, A.W. Dahmani2, F. Askri3, M. Ben

Salah3 and S. Ben Nasrallah4, 2009, “Experimental study of a

parabolic solar concentrator,” Revue des Energies

Renouvelables Vol. 12 No. 3, pp. 395 – 404.

[4] C. Lertsatitthanakorn1, J. Jamradloedluk1 and M.

Rungsiyopas2, 2014, “Electricity generation from a solar

parabolic concentrator coupled to a thermoelectric module,”

2013 International Conference on Alternative Energy in

Developing Countries and Emerging Economies, Energy

Procedia 52 , pp. 150 – 158.

[5] Lifang Li1, Steven Dubowsky, 2011, “A New Design

Approach for Solar Concentrating Parabolic Dish Based on

Optimized Flexible Petals,” Mechanism and Machine Theory

46, pp. 1536–1548.

[6] Hanlon, J., 1992, “1st ed. Handbook of Package

Engineering”, ISBN 0-87762-924-2. Chapter 3 Films and Foils.

http://scienceworld.scholastic.com/Physics-

News/2012/12/optics.


Proceedings of the Fifteenth Annual Early Career Technical Conference The University of Alabama, Birmingham ECTC 2015


APPLICATION OF A SEQUENCE OF DESIGN METHODOLOGIES TO THE PROBLEM OF TRANSPORTING WARM PARTICLES IN PARTICLE HEATING RECEIVER SOLAR ENERGY

SYSTEMS

Kenzo Repole, Sheldon Jeter

Georgia Institute of Technology Atlanta, GA, USA

ABSTRACT The central receiver power tower (CRPT) with a particle

heating receiver (PHR) is a form of concentrating solar power

(CSP) system which has the ability to achieve high efficiency at

low cost and to readily incorporate thermal energy storage

(TES).

A critical component in such a PHR system is the particle

lift system, which must transport the particulate from the lower

temperature TES bin back to the PHR. This particle lift system

is a critical, innovative solution drawing from many industries,

such as the mining industry, yet it is being qualified through a

special sequence of design methodologies in order to develop a

viable design neutral solution.

INTRODUCTION As countries around the world move towards increased use

of renewable energy and to find ways of climate change

mitigation, governments have started to institute stricter

regulations to reduce the use of fossil fuel and to encourage

renewable energy sources. One such set of regulations has been

issued by the Environmental Protection Agency in the United

States [1].

These types of policies mean that renewable energy will

play a greater role in delivering electricity via the power grids.

However, to ensure dependable base loads either a fossil fuel

plant must be used to generate electricity or some form of

storage must be used to access energy from renewable sources

when they are unavailable.

These forms of energy storage can range from chemical

batteries, flywheels, compressed air, capacitors, or thermal

energy storage [2].

One form of solar energy power is the concentrating solar

power (CSP) system. A particular form of CSP uses particles to

capture the solar energy then convert it to electricity. This is

known as a particle heating receiver (PHR) system. This system

has the ability to incorporate economical thermal energy

storage (TES).

In a PHR system, solar energy is concentrated from a field

of heliostats onto a falling curtain or layer of particles at the

receiver aperture of a power tower. [3]

The particles can then be directed towards a heat exchanger

in a power generation system or to storage for later use when

the utility load demand is greater or as a means to compensate

for the solar variation experienced during the day.

The advantage that PHR systems have over other forms of

CSP is the ability use low cost materials to store sensible

energy over a longer period of time. Storage allows off-sun and

night time generation, and extended use of the energy plant

reduces the levelized cost of energy (LCOE) [4].

As the need for greater capacity CSP, especially PHR

systems, grows so will the need to find an effective delivery

system to recycle the working particles to the PHR at faster

rates and in larger amounts.

Larger capacity PHR systems with TES means the power

tower will increase in width and especially in height in order to

store the high temperature particles used during power

generation or as storage for demand based power generation.

Currently, there are many different candidate methods of

carrying fine particles up to elevations of the PHR. However, to

increase the capacity of the power plant and its efficiency, the

particles entering the PHR need to be at higher temperatures

ranging from 300°C (572°F) up to 600°C (1112°F). At such

temperatures some conventional methods of delivering large

amounts of working particles are not viable, since the operating

environment is outside the operating range for some delivery

mechanisms. [5]

This paper outlines the processes used to develop candidate

designs, and the further integration of the favored conceptual

design is discussed. The process is the application of a different

set of design methodologies used in a special sequence in order

to elicit a feasible design solution from vague requirements. It

is felt that this special sequence could be used in similar

situations.

This design is especially demanding, since it is outside

what is currently used in commercial or industrial applications.

The final solution would need high reliability in elevated

temperatures, where a lift of tons of fine particles over

hundreds of meters is needed.

The final solution is not refining an existing design but

finding a new solution for an application never used before.

Just as importantly, the final solution is not meant to be

mass produced and but to be efficient and cost effective where

many iterations and extended research are not feasible.

DESIGN METHODOLOGIES There exist many design methodologies for concept

generation. However, they can be mostly grouped into three

types. These are the conventional methods, intuitive methods

and the discursive or systematic methods [6].


Conventional methods encompass such things as patent

and literature reviews, analysis of existing systems, analyzing

natural systems or using analogies to formulate a design [7].

These processes are best suited for product improvement and

mass production products.

Intuitive methods include processes like brainstorming [8],

mind mapping [9], or the 635 Method [10]. These processes are

better suited for innovative or inventive products.

The systematic methods are design-neutral methods, which

follow a step by step sequence in order to come to a best

solution. Such processes are like the Morphological chart

method [11] or Axiomatic design [12]. These processes are

meant to have a design neutral analysis giving allowance for

intuitive design or conventional methods.

THE PUGH METHOD When designing solutions for novel needs or unique use, it

is helpful to use a best practice design methodology. Such a

methodology helps to account for important criteria, which

could be overlooked, and encourages a design neutral process

to develop a fully comprehensive design.

One of the best practices in design methodology is the so-

called Pugh concept selection [13]. After the customer needs

have been compiled and the initial round of function

requirements and selection criteria developed, different

concepts are generated based on this information. In the Pugh

concept selection process, one concept generated is considered

to be the reference design, and all other designs are compared

using the equally-weighted selection criteria. The scores are

then used to rank each design. The top ranked designs can be

compared a second time using unevenly weighted criteria to

focus on the most suited design solution. Once the highest

ranked conceptual design solution is selected, it can then be

developed into a more detailed design.

AXIOMATIC METHOD A best practice design methodology to evaluate concepts is

the axiomatic design method [27]. In this process, functional

requirements (FR) in the physical domain are developed from

customer needs and then mapped to design parameters (DP) in

the design domain. The DP is the feature required to meet the

FR.

Once this mapping is done, the design parameters become

the basis for development of more detailed FR, and this next

level of FR becomes the basis for the next set of more detailed

DP. This process is repeated until the functional requirements

and design parameters have reached a sufficiently detailed

level.

After this definition is achieved, the Independence Axiom

and Information Axiom [27] can be applied to the FR and DP in

order to determine if the design is a suitable design in the sense

of being robust with respect to design interactions. For the

independence axiom, this is normally illustrated in a matrix

form as {FR} = [ A ]{DP} where the design matrix is [ A ]. The

ideal design in this sense would result in a design matrix that is

diagonal. This means that each FR is independent of the other

FRs. Such a design is called an uncoupled design. The design

can easily be optimized since each FR can be modified

independently of the other FR.

If the design matrix closely resembles a triangular matrix

this design is considered a decoupled design. Most designs are

decoupled, since many of the different parts of the design

architecture depend on each other at interfaces.

The least desirable design is a coupled design. This would

result in a full (or nearly full) design matrix. This would

indicate that every FR is linked to all other FR. This design

would be very difficult to optimize, and it would be difficult to

change any FR without having to modify all other FRs.

DESIGN FOR MANUFACTURING Design for manufacturing is a process whereby a product

design is adjusted in order to increase the ease with which it

can be manufactured and assembled. Such aspects of the design

would be to use reduced cost materials, standardized parts, and

to use dimensions compatible with available transportation

methods or to set the shape of components for ease of

manufacturability or assembly. Also, a modular design would

be adopted with the reduction in the number of parts in order to

lower cost and increase the efficiency of assembly [14].

CUSTOMER NEEDS As part of the US Department of Energy’s the Sun Shot

Program's project for the "Development of a High Temperature

Falling Particle Receiver" a means for transporting particles to

the top of the CSP tower was needed. However, the exact

specification were not given; only that the lift must maintain

particles being transported close to 300°C (572°F) and must

deliver the particles in a manner to ensure a constant mass flow

rate with respect to thermal energy requirements. As the design

process evolved more functional requirements and design

specifications were developed.

CONCEPT GENERATION AND SCREENING Due to time and budgetary constraints, it was decided to

use a sequence of different design methodologies in order to

arrive at a viable design solution. Since the exact customer

needs were not specified, axiomatic design elicitation was used

to develop the first level of functional requirements.

This first level of functional requirements was then used as

the initial selection criterion during the conventional method of

patent and literature review. This review was conducted to

understand the current solutions for transporting high

temperature particles on an industrial scale within the area of

the first level of functional requirements.

After this the qualified options where screened and

compared to a reference solution which was used in a previous

prototype. The most viable option was then put through the

intuitive design process of mind mapping in order to generate

any other requirements which were not readily apparent.


FUNCTIONAL REQUIREMENTS For our initial design, a particle transport solution was

needed to deliver particulates up to a height of 138 m of a 60

MWth solar power tower, after which we applied our findings

to a proposed commercial scale 100 MWth solar power tower.

The height of the lift is determined by the size of the high

temperature bin used to store the particulate for use during the

day and the assumed 9 hour off-sun period. Based on this

information and other constraints, the first level of FR was

developed and is shown in Table 1.

Table 1 First Level Functional Requirements

FR# FR Description

FR01 Be able to transport vertically large amount of small

particulates

FR02 Be able to operate in a shaft temperature environment

between 150 to 200°C

FR03 Be able to operate with minimal heat loss

FR04 Be able to operate with minimum particulate spillage.

FR05 Be able to resist wear regardless of size or hardness of

particles

FR06 Be able to have an overall energy efficiency of 75%

FR07 Be able to have dimensions that allow transportation

on rail and shipping containers

FR08 Be able to meet safety factor minimum for industry

related standard.

It was then decided to use these FR as selection criteria for

the initial round of concept generation and selection using the

Pugh concept selection.

PARTICLE TRANSPORTATION OPTIONS Some of the options that are available to meet this

challenge are bucket elevators, Olds Elevators, conveyer belts,

and particle skips similar to those used in the mining industry.

Suitable bucket elevators may have the ability operate at

high temperatures, greater than 200°C (392°F) [15]. However,

this technology would typically require the shaft for particle

transport to be kept at the same temperature as the particles for

a reasonable heat loss. Moreover, this system would likely

experience a high spillage rate during operation.

Olds elevators (OLDS) [18] have the ability to deliver the

working particles in a continuous flow and high temperature.

However, as the height of the tower increases the cost of the

OLDS increases significantly due to the nature of its design.

Nevertheless, the OLDS was used as the reference solution

since it is used in the known PHR prototypes [19].

Conveyer belts have little spillage but are difficult to

integrate into a tower and are not suited to convey high

temperature particles without huge heat loss. Application of the

Pugh method as seen in Table 2 shows that the most suitable

design option that can address the current and future needs of

larger capacity PHR that maintain high thermal efficiency and

low exergy degradation is the particle lift.

SKIP LIFT ALTERNATIVES

Particle lifts similar to those used in the mining industry

come in different forms. The main forms currently used in the

mining industry are Bottom Dump skips, Front Dump skips,

Kimberly Skips (KS), and Arc Gate skips.

The main tradeoffs between the different skips types are

(1) the extra height required during operation, (2) ease of

operation at high temperature, (3) spillage during use, (4)

maintenance requirements, and (5) simplicity of effective

thermal insulation.

Bottom Dump skips are charged from the top and

discharged by a trap door forming part of the bottom of the

skip. It does not require a large amount of extra height for its

operation in comparison to the other types of skips. This skip

design is light weight and rugged; but due to the fine size, 250

nm, of the particles used in the PHR, spillage may be large

during the transport and discharge of the particles.

Front Dump skips, are charged from the top and discharged

through a gate forming part of the lower end of the front side of

the skip. Such skips are reportedly able to carry large volumes

of particles and to put the least amount of stress on the head

frame [20]. However, the spillage rate may still be high in

comparison to other types of skips.

Arc Gate skips are considered to be safe and rugged [20].

They are charged from the side and discharged through an arc

gate on the side. As with the bottom dump skip, the Arc Gate

skip may experience large amounts of spillage during charge

and discharge. In addition to this skip has many moving parts

implying an increased risk of failure under high temperature

and extreme environments that would be experienced in

transporting the working particles in the PHR.

Kimberly skips, are charged and discharged from the top of

the skip. The particles are loaded into the skip from the top. The

skip then travels in this configuration until it reaches its

discharge location. As it reaches the discharge location, a set of

scroll wheels on the skip engages scroll guides on the shaft

walls. These wheels guide the skip through the dump zone and

allow the skip to rotate to about 120° from its vertical position.

This action discharges the particles from the skip. The KS is

expected to have the lowest initial cost, the lowest maintenance

cost and highest service life in comparison to the other skip

types [20]. KS have the lowest amount of spillage occurrence

during use. However, KS requires larger headroom and width

clearance than any other skip design. They also exert the largest

amount of stress on the head frame [21], [22].

In our development, the Front dump and Kimberly skips

appeared to be most promising, so scale models were

developed for qualitative comparison. Operation of the scale

models indicated that the KS would be easiest to effectively

insulate and operate at the expected high temperatures.

Another important subsystem in the particle lift is the hoist.

There are two main types of hoist [22]. The drum type has a

single rope or multiple ropes wound on a drum and controlled

by an AC drive motor. The other main type is the Koepe

friction hoist. In this design, two skips or one skip and a

counter balance are operated by ropes passing over a drum, and


the friction between the ropes and the drum controls the motion

of the skip.

The Koepe hoist is reported to be the most common hoist

system used in the mining industry today. It is based on using

the friction between the drum and the wire rope to enable the

drum to drive the skip operation. Despite its attractive low cost

and wide use, it was not initially selected in this project due to

concerns about reliable friction-dependent high temperature

operation, the reason being there is a possibility that the heat

generated from the friction in addition the high temperature of

the shaft could greatly reduce the life of the wire rope in use.

Therefore for this design a Blair [22] drum hoist was

selected. One of the advantages of the Blair drum is its ability

to run the skips independently of each other in cases of

emergency, thus giving the system a contingent means to

continue operating if one skip becomes nonfunctional.

REFINING PARTICLE LIFT SELECTION After selecting the general particle lift concept by the Pugh

method and generating a more detailed level of DP by mind

mapping as seen in Figure 1, the design problem was further

analyzed using the Independence Axiom. For this analysis the

more detailed or so-called “drill down” FR are shown in

Table 3, and the corresponding drill down DP are shown in

Table 4.

Table 2 Pugh Matrix for the Concept Selection

FR

Options

Bucket

Elevator

Olds Elevator

(Reference)

Conveyer

Belt Lift Hoist

System

FR01 0 0 0 0

FR02 0 0 - -

FR03 - 0 - +

FR04 - 0 - +

FR05 - 0 0 +

FR06 0 0 0 0

FR07 0 0 0 0

FR08 0 0 0 0

Sum + 0 0 0 3

Sum 0 5 8 5 4

Sum - 3 0 3 1

Net Score

-3 0 -3 2

Rank 3 2 3 1

Figure 1 Mind Map for Particle Lift Solution


Table 3 Detailed Functional Requirements

FR# Description

FR01-01 Be able to transport 250 nm size particulates

FR01-02 Be able to transport particulates up to 140 m

FR01-03 Be able to transport 20000 kg of particulates within 60 seconds

FR02-01 Material must be able to maintain desired strength

between 150 to 200°C environmental temperature

FR02-02 Fluids used including lubricants must be able to maintain

desired properties between 150 to 200°C

FR03-01 Skip must be insulated

FR04-01 Overall system must have less than 1% spillage or

system for recovery

FR05-01 Impact surfaces need to resist wear from loading and unloading particulates

FR05-02 All moving parts and ropes need to resist abrasion wear

either by clearance or debris removal mechanisms

FR06-01 Be able to have an overall energy efficiency of 75%

FR07-01 Skip maximum dimension must not exceed

LxWxH (2 m x 2m x 12m)

FR08-01 Single Rope Factor of Safety 5 [11]

FR08-02 Rope nominal diameter cannot exceed 76 mm

FR08-03 Rope contains independent core rope that can withstand shaft temperatures

FR08-04 Overall system should have factor of 5 [16], [17]

Initially, several lift options were considered. Analysis

identified the counterbalanced Blair drum hoist as the most

promising based on efficiency, cost, and reliability. Two

generic skip types were considered most promising: (1) the

Front Dump Skip and (2) the Kimberly Skip. The Front Dump

Skip is evidently favored in traditional mining, since its layout

is compatible with a relatively small cross section and longer

length. The smaller cross section is highly desirable in mining

where the vertical shaft can be hundreds to thousands of meters

deep. In contrast, the simplicity of the KS promotes a low

initial cost, low maintenance cost and high service life. All

these features are important in CSP applications; therefore the

KS was selected for this application.

The qualitative analysis (including construction and

operation of two scale models) outlined above identified the KS

with Blair drum hoist as the promising design. Furthermore

application of Axiomatic analysis shows that this combination

is a highly suitable design but not an ideal design. The resulting

design matrix, as seen in Figure 4 , is close to a triangular

matrix, meaning the design is largely decoupled. There was

clustering around FR for rope selection and temperature, as

expected. This indicates that these two criteria are critical to

success of this design.

Table 4 Detailed Design Parameters

DP# Description

DP01-01 Kimberly Skip Design

DP01-02 Blair Drum Type

DP01-03 AC Electric Drive Motors

DP02-01 Metal for Skip is of SS 316-H material 3 mm thick

DP02-02 Lubricant with NLGI No 2 and flash point over 200°C

DP03-01 Fire Brick with Thermal Conductivity of 0.32 W/m-K @

800°C and Density 800 kg/m3

DP04-01 Olds Elevator in Sump to remove spillage

DP05-01 Loading and Unloading angles greater than 20 degrees

DP05-02 Bearing and joint material with Hardness greater than

ID-50K

DP06-01 Lift efficiency greater than 75% & Recovery Efficiency

greater than 75%

DP07-01 Skip dimensions LxWxH (1m x 2m x 4m)

DP08-01 Rope Safety Factor of 5

DP08-02 Rope Diameter between 31.7 mm and 50.8 mm

DP08-03 Rope Core is SS316 material

DP08-04 Rope Safety Factor greater than 5 & Skip Safety Factor

greater than 5

All such spillage will be accumulated in a sump built into

the lift shaft, which can be emptied as necessary; therefore,

there will essentially zero net loss of particulate from the

system. Minimal heat leak is also an objective, and simplicity

of the proposed skip design makes it easy and inexpensive to

install adequate internal insulation to keep the heat leak from

the skip well under 0.1% of the rated capacity of the system.

Our design also envisions a lift shaft allowed to stay at

200°C (392°F), which further minimizes incidental heat leaks.

Altogether the proposed design ensures a minimal heat leak that

will have negligible effect on the overall system efficiency.


No. Name

1 Lift Machine Room

2 Lift Discharge Chute

3 Particle receiver

4 High Temperature TES Bin

5 PWF Heat Exchanger

6 Low Temperature TES Bin

7 Lift Charge Chute

8 Lift Shaft

9 Top hopper

Figure 2 Lift integrated into TES Tower. [5]

Figure 3 Conceptual Insulated Kimberly skip charging, travel position and discharging [5]


=

Figure 4 Independence Axiom Matrix Analysis for Kimberly Skip with Blair Drum

COMMERCIAL SCALED PARTICLE LIFT The analysis and concept selection described above

resulted in the selection of KS with Blair hoist. Our next task

was to develop this concept into design for a commercial

particle lift that would be part of the 460 MW-th CSP tower as

seen in Figure 2. The design and engineering of a commercial

particle lift system was then completed. Conceptual design

drawings and energy efficiency and heat loss modeling have

also been completed. Detailed efficiency modeling based on

reliable published component efficiencies resulted in an energy

efficiency of 80%, which exceeds the 75% energy efficiency

metric. With this efficiency, the parasitic power should be less

than 1% of the rated output. The selected skip design is that of

the KS type, seen in Figure 3 with its charging, travel and

discharge configurations.

This skip is both filled and discharged from the top and

does not have a complicated and leak-prone bottom hatch. This

arrangement facilitates a design that is very simple structurally

and mechanically. The single top hatch, which is opened and

closed by motion of the skip, thereby eliminating any

mechanical or hydraulic actuators, is critical to this simplicity.

Importantly, this design appears to be almost leak proof and

should easily achieve much less than 0.1% temporary spillage

of particulate during filling and discharge.

The first set of drawings and specifications has been

completed, and we have already consulted with one company

familiar with steel fabrication and industrial lift manufacture.

This company has commented that our design will be easy to

fabricate. After incorporating some minor modifications based

on this review, we also consulted informally with a skip-hoist

component supplier. With helpful input from these initial

reviews from smaller companies, we contacted two of the

major manufacturers.

One of these manufacturers has commented that our

design should be generally feasible to fabricate and install;

however, they have responded that the Koepe hoist and a

bottom discharge skip should also be considered. The Koepe

hoist would probably be simpler and less expensive, and it

should also have lower drum inertia, which would reduce

dynamic loads and deceleration losses. Going forward, we will

definitely be considering these alternatives.

The simple design (basically a bucket with a hinged lid and

a lifting bail) allows effective thermal insulation with mere

layers of continuous suitable rigid insulation such as firebrick

inside the skip and the lid with no complicated bottom hatch to

insulate and no mechanism components (such as links and

latches) to act as thermal short circuits. The bottom hatch is

also likely to leak during lifting, which is not an issue when

handling typical raw materials but which is important when

hoisting the TES medium. Our experience with the two small-

scale models was convincing with regard to these issues.

As shown Figure 2 the Kimberly skip is easy to integrate

into the CSP system. Note that the lift shaft will be kept at

elevated temperature around 200°C (392°F) to minimize heat

losses, but the electrical and mechanical equipment (other than

the lift drum) will be kept at near ambient temperature for

efficiency and economy.

Typically, stainless steel such as SS316 wire rope is

selected for durability, corrosion resistance, and excellent high

temperature strength. Cost estimates based on a highly regarded

source of generic unit and subsystem costs are summarized in

the following Table 6.

MODIFICATION OF DESIGN FOR MANFACTURABILITY The design was then modified to make it more suitable for

manufacturability. For example the LxW dimensions were

made equivalent in order to fit inside shipping containers and

on trains. This also allowed the volume to change, thus

reducing the height needed.

Also, the safety factor was relaxed to 3 since the design

was considered as an industrial application where personnel

would be not be present in the vicinity during operation. This


modification has allowed the reduction of wire rope diameter to

below or equal to 76 mm (3 inches). In general, 76 mm is the

upper limit for easily available SS316 wire rope. The modified

design specification is shown in Table 5.

Table 5 Design Specifications for Modified Particle Lift

Design Specification Value

Power Capacity of Tower 460 MWth

System Mass Flow 979 kg/s

Skip installed 2 skips

Estimated Skip Dimensions (LxWxH) 2m x 2m x 6m

Ropes in use per skip 1 rope per skip

Rope Type SS 316 6x37 IWRC

Rope Diameter 76 mm (3 in.)

Drum to Rope Diameter Ratio 60

Rope Layers on Drum 3

Rope wraps per layer 5

Electric Motor AC Induction Motor

Gear Reduction Ratio 46

Overall Safety Factor 3

Table 6 Current Cost Analysis for Particle Lift

Estimated Cost of 2 Skips without Hoist System $198,533

Estimated Cost of Hoist System $294,600

OLDS Elevator Particulate Recovery system $30,000

Total Estimated Cost per System $523,133

Total Estimated Particle Lift cost per MWth $8,719

The cost analysis was conducted based on design of the

skip and using typical values found in cost databases for other

necessary parts of the system including the drive, control and

guidance systems[23].

The total estimated cost per system was determined to be

$523,133; this was then compared to the costing analysis

methods developed by Sayadi [24] which was $538,567 and

typical values found in the mining industry [22], which was

$1,171,900, all in US Dollars (2014).

The cost analysis and the estimate using the method

developed by Sayadi were close in value; however, the value

normally found in industry [22] was almost double this value. It

is felt that the difference is due to the overhead charged by

hoist vendors included in the hoist installations including

installation consulting fees, maintenance service agreements

and multi-year license and support agreements for propriety

software used in lift control systems. The estimates did not

include the construction of the lift shaft.

Accordingly, the lift system is expected to cost around

$8,719 per MWth, which agrees with previous independent cost

estimates using technology-specific cost engineering research

results, which was around $9,614 per MWth [25].

This analysis was conducted assuming a module of 60

MWth. A system size up to 460 MWth would be

accommodated by some combination of larger skips and

multiple pairs of skips.

Typically, costs per unit are improved at larger size, and

overall efficiency is only negligibly changed with larger system

size and longer lift.

The rope size of 0.076 m (3 inch) based on the above

calculated stress and on vendor tensile strength of SS316 using

the factor of safety of 5 as required by OSHA[26] for wire rope

and many other critical applications. A multi-rope system may

need to be considered in future analysis, since smaller rope

diameters are easier to source and are more flexible.

The energy efficiency modeling, which is still being

refined, is based on lift and recovery efficiencies and the ratio

of overall tare to payload in Table 7. The tare fraction is

important, since the potential energy of the skip and rope

cannot be 100% recovered. Also, the lift efficiency of 85% is

from several published standards and models and has been

confirmed by component modeling shown in Table 8. The

expected energy efficiency is 80%, which is higher than the

target of 75%.

Table 7 Estimates of overall efficiency for particle lift design.

Data Efficiency

Lift Efficiency 0.85

Recovery Efficiency 0.93

Ratio: Tare/PL 0.2549

Overall Efficiency 0.8064

Fraction Parasitic 0.0086

Table 8 Estimates of lift component efficiency for particle lift design.

Component Efficiency

VF Drive 0.96

Electric Motor 0.95

Gearing, 2 Stage 0.98 x 0.99 = 0.97

Rope/Drum Efficiency 0.98

Overall Product 0.86 to 0.87

Some remaining more detailed aspects that will be

investigated in future include tribological analysis and design

and material selection for all bearings, joints, and ropes. In

particular, situations where similar materials are in contact can

result in types of adhesive wear, such as galling or scuffing.

Furthermore the metals are typically softer than the particulate;

therefore, the possibility of third body abrasive wear [28] is

significant. The final design must ensure that such tribological

effects do not reduce the life of the skip or hoist system.

FURTHER WORK Currently, plans are under way to build a scaled working

model of the particle lift and to refine the cost model to reflect

more accurately the particle lift components and to account for

vendor markups. More importantly, research is planned to


verify outstanding issues with elevated temperature effects on

lubricated parts and material endurance on such components as

the wire rope.

CONCLUSIONS In conclusion, as the need for better performing and larger

capacity CSPs increases, the need for lower cost and higher

capacity TES will follow suit. This will be especially true for

solid particle based TES.

It has been shown that out of the many different systems

that can be used to transport the particles up to the PHR, a

suitable commercial solution would be a KS based particle lift.

This design would have low cost per unit of CSP power

capacity, high overall efficiency, long service life, and low

maintenance cost with proper tribological design. More

importantly, the sequence of design processes used developed a

viable solution which could easily be implemented in a

commercial project scope and other similar research projects

could use this exact sequence to bring forth a solution within

the scope of the project with time and budget constraints.

ACKNOWLEDGMENT Part of this work was supported by the US Department of

Energy through the Sun Shot Program's project for the

"Development of a High Temperature Falling Particle

Receiver" (Project ID: DE-AC04-94L85000). The Prime

Contractor is Sandia National Laboratories, and the Sandia’s PI

is Dr. Clifford K Ho. The financial and programmatic support is

recognized and greatly appreciated

REFERENCES [1] EPA” Clean Power Plan Final Rule“ 40 CFR PART 60

(2015) Docket No. EPA-HQ-OAR-2013-0602

[2] Denholm, P., Ela, E., Kirby, B. Milligan M. “The Role of

Energy Storage with Renewable Electricity Generation”

NREL/TP-6A2-47187, January 2010

[3] C. Ho, J. Christian, D. Gill, A. Moya, S. Jeter, S. Abdel-

Khalik, D. Sadowski, N. Siegel, H. Al-Ansary, L. Amsbeck, B.

Gobereit, R. Buck, Technology Advancements for Next

Generation Falling Particle Receivers, Energy Procedia,

Volume 49, 2014, Pages 398-407.

[4] Ma Z, Glatzmaier G, Mehos M, Fludized Bed Technology

for Concentrating Solar Power with Thermal Energy Storage.

Journal of Solar Engineering 2014; 136

[5] Repole, K., Jeter, S., (2014) “Preliminary Commercial

Design For Transporting Low Temperature Particles for Use in

Particle Solar Receivers Using Lifting Skips “Proceedings,

Fourteenth Annual Early Career Technical Conference,

Birmingham, AL, page 63.

[6] Pahl, ., Ken allace, and Luci nne Blessing. Engineering

Design: A Systematic Approach. London: Springer, 2007. Print.

[7] Fu, K., Murphy, J., Yang, M., Otto, K., Jensen, D., and

Wood, K., 2015, "Design-by-analogy: experimental evaluation

of a functional analogy search methodology for concept

generation improvement”, Research in Engineering Design, 26

(1), 77-95.

[8] Pahl, .; Beelich, K. H.” Lagebericht. Erfahrungen mit dem

methodischen Konstruieren.”Werkstatt und Betrieb 114 (1981)

77381) 7

[10] Rohrbach, B.”Kreativ nach Regeln eMethode 635, eine

neue Technik zum Losen von Problemen.”Absatzwirtschaft 12

(1969) 73aft.

[9] Beel, J., Langer, S.,.” An Exploratory Analysis of Mind

Maps.” In Proceedings of the 11th ACM Symposium on

Document Engineering (DocEng’11), Mountain View,

California, USA, pages 81-84 2011. ACM.

[11] Seghedin, N. E., & Chitariu, D. F. (2014). “Software

System For The Development Of Morphological Matrixes Used

In Technical Creation”. Elearning & Software For Education,

(4), 377. doi:10.12753/2066-026X-14-289

[12] Thompson, M. K., “Improving the requirements process in

Axiomatic Design Theory”, CIRP Annals - Manufacturing

Technology, Volume 62, Issue 1, 2013, Pages 115-118, ISSN

0007-8506, http://dx.doi.org/10.1016/j.cirp.2013.03.114

[13] Ulrich, Karl T., and Steven D. Eppinger. Product Design

and Development. Boston: McGraw-Hill/Irwin, 2004. Print.

[14] Whitney, D. E. (1988). “Manufacturing by design.”

Harvard Business Review, 66(4), 83-91.

[15] Rexnord Corporation

[16] Shigley, Joseph Edward., and Charles R. Mischke.

Mechanical Engineering Design. New York: McGraw-Hill,

1989. Print.

[17] CFR Title 30, Part 57.19019

[18] Olds Elevator LLC: http://www.oldsusa.com

[19] Golob, M., Abdel-Khalik,S.,Gill, R.,Jeter, S., Nguyen, C.,

"Solar Simulator Test Of Wire Mesh Particle Heating Receiver

To Measure Receiver Efficiency" Proceedings, Fourteenth

Annual Early Career Technical Conference, Birmingham, AL,

Paper 56

[20] "Ore Skips, Mine Haulage Systems and Other Foundry

Products in Ontario." Wabi Corp. N.p., n.d. Web. .

[21] De la Vergene, J., “Hard Rock Miner’s Handbook”,

Edition 3, 2003, McIntosh Engineering, Ontario, Canada.

[22] “SME Mining Engineering Handbook”, Third Edition,

2011, Society for Mining, Metallurgy, and Exploration Inc.

[23] RSMeans: http://www.rsmeans.com

[24] Sayadi AR, Lashgari A, Oraee K, Yavari M (2010)

Hoisting Equipment Cost Estimation in Underground Mines.

SME International Conference on Hoisting and Haulage, Las

Vegas, USA.

[25] Research Performance Progress Report (RPPR), High

Temperature Falling Particle Receiver, DE-EE0000595-1558,

15 July 2013.

[26] OSHA “Safety and Health Regulations for Construction”

29 CFR 1926.251(2010)

[27] Suh, N. P. (2005). Complexity: Theory and applications.

New York: Oxford University Press.

[28] Williams, J. A. Engineering Tribology. Oxford: Oxford UP,

1994. Print.

http://dx.doi.org/10.1016/j.cirp.2013.03.114

http://www.oldsusa.com/

http://www.rsmeans.com/




ANALYSIS OF AN EXPERIMENTAL BUILDING HVAC SYSTEM TO IMPROVE EFFICIENCY

Christopher Fernandez


Sheldon Jeter


ABSTRACT

The goal of this research project is to develop an easily

adaptable simulation model that represented a generic multi-

purpose building in higher education and research to determine

what heating, cooling, and ventilation systems are the most

energy efficient. To have full control of the building, a

computer based simulation software is used to adjust

parameters within a realistic budget. As such, some

assumptions are made, such as that the weekday people and

lighting loads are equal, to make scheduling loads easier and

consistent. While electrical usage is monitored, other

parameters are also factored into the model, such as: chilled

water thermal energy, hot water thermal energy, room air

changes per hour, occupant comfort, and contamination in the

form of high humidity and carbon-dioxide. It is found that

situational cost analysis based on electrical cost compared

favorably to access to supplied cooling and heating sources.

INTRODUCTION It is well known that the majority of electrical power

consumed in the United States by buildings goes into climate

control. With the rise of energy prices coupled with an

emphasis on renewable and efficient energy, developing a new

way to insure personal climate comfort and low energy usage is

a growing concern.

The objective of this work is to run simulations for a

building that represents a large majority of buildings on the

Georgia Tech campus. This required that the model building

have no unique architectural features or complications. The

experimental building is designed to be a typical low-rise

building with three floors and 3 zones per floor; the zones can

be changed to suit any design but are left as offices for

simplicity. Being an education-focused building, the lighting

and plug loads are set higher than a typical office building, but

can easily be changed to reflect conditions in any building.

Most buildings also have a perimeter zone and offices in

interior zones, but most buildings on the Georgia Tech campus

have exterior zones and internal passageways to allow labs and

classrooms to have windows. The flexibility coupled with ease

of schedule and zone changes allows a quick adaptation of this

model to most any building in higher education. While day to

day people, plug, and lighting loads realistically change, for the

sake of convenient schedules and to avoid different loads on

different simulations, the weekend and weekday schedules are

kept constant throughout the simulation year. In addition,

natural inefficiencies i.e. open windows, leaking ducts, etc. are

not included to insure all energy changes are the result of the

HVAC system and not a reduction in losses.

This paper discusses the details of the model, including

why certain design considerations are made. Software played a

major role in the research and development; the benefits and

problems that are overcome are detailed as well for those

unfamiliar with the programs used. In addition, some basic

explanations on how the different softwares go about

thermodynamic, contaminant balance, and heat transfer are also

provided. A desire to keep only a small number of variables in

the simulation is another priority in development, so a

procedure is established to give consistent results for every

simulation. The details of this are presented before results are

given so that one can understand the process before discussing

results of said designs.

MODEL DETAILS The model is broken up into three floors with three zones

per floor, as in Figure 1: the longer Main zone with Side A and

Side B having the same design as each other. The design is

based on most academic buildings on the Georgia Tech campus,

where corridors are small and in the interior, while offices and

classrooms are placed along the perimeter. The model building

included minimal fenestration along the West facing

fenestration; the majority of the fenestration is located on the

South and North faces, ensuring that all rooms have an ample

view. There is no fenestration on the East wall. While excluding

an interior zone may seem like a flaw in design, the people load

in a hallway is typically low and is an artifact of people

entering and leaving the individual zones. As such, a constant

supply of air could meet the thermal load of a hallway in

conjunction with substantial thermal capacity. In addition, one

could establish the building to have return ducts in the hallway

so that it would be conditioned by the air leaving the zones.

Having separate, smaller zones allows for greater zone control

over parameters and loads. Just as in a real building, all the

zones are not used equally throughout the day, structuring the

building to allow for fluctuations in usage, which are

independent.


Figure 1. Top down view of zone layout

Figure 2. Georgia Tech Whitehead Building

While energy conservation is the main objective, having

occupants feel comfortable with large windows to the exterior

is incorporated into the design in Figure 3. Some modern

design features are incorporated in Figure 4, such as minimal

fenestration on the West and South, longer along the East/West

direction, high North facing fenestration as seen in the

Whitehead Building in Figure 2. [1].

Figure 3. SketchUp view of South and East walls

While the physical shape of the building is not changed

through the modeling iterations, the methods of meeting the

thermal loads and outdoor air specifications are adjusted and

compared to each other. With these criteria established, the

objective of finding the most energy efficient building is

pursued.

Figure 4. SketchUp view of South and East walls

SOFTWARE

EnergyPlus is a program provided by the US Department

of Energy; it is designed for complete building simulation of

energy, people, utility usage, lighting, energy flows, and many

other features. The program also allows users to establish how

rigorous the simulation should be. The program has an

abundance of beneficial abilities, such as automatic conversion

between English and Metric units, converting material

properties into constructions for boundaries, auto calculating

sizing and heat loads, and the ability to use real world weather

data.

EnergyPlus is an ideal simulation tool for buildings

because it is a combination of a thermodynamic, mass transfer,

and concentration model. The benefit is that these three

parameters can be adjusted independently of each other yet can

still function together. In addition, there are a number of

simulation parameters that can bring EnergyPlus to an all-

inclusive level of simulation and customization. There are

options for different convection and conduction algorithms that

apply to different methods of building modeling and heat

transfer levels.

Practically any simulation and method for heat or particle

transfer can be represented through EnergyPlus. At its most

advanced, it is even capable of monitoring moisture diffusion

and evaporation through layers of materials depending on

material properties and temperature gradients, isotherms, and

conductivity.

EnergyPlus is also compatible with other software such as

Google SketchUp and OpenStudio, which are used to create the

building geometry. Other programs are used for geometric

construction because EnergyPlus relies on a coordinate system

input for walls, floors, roofs, and fenestrations while Google

SketchUp uses a graphical interphase.

In addition to external software compatibility, EnergyPlus

allows for external data entry directly into the simulation. This

includes, but is not limited to, actual electrical loads, which can

be reduced into specific loads per zone, occupancy loads for

individual zones, and real world weather. While not necessarily

required for a simulation, understanding how a building

realistically is used and the weather conditions it will be subject

to can greatly influence how to properly setup a building to

reflect actual usage. While EnergyPlus includes TMY2 weather

data, weather data that reflects an average over the entire year,

it is possible to gather actual weather information and simulate


the building that way. While it is not recommended to design a

building based on only one year of data, this allows one to see

how a stretch of extreme weather could overload a HVAC

system. On the Georgia Tech campus, a local weather station is

setup to record wet and dry bulb temperatures, humidity, solar

irradiance over the horizontal and direct, as well as rain, wind

speed and direction.

PROCEDURE The building structure and layout in Figure 5 was created

in Google SketchUp and has a design of a building that

represents a majority of the buildings on the Georgia Tech

Campus. No hallways are added, so that the entire building

would be used as occupied space; this is done for simplicity.

While this may lead to an inefficient design where the envelope

of the building directly affects the thermal comfort of the

occupied zone, it also makes changes to the model more

noticeable. In addition, most hallways are in the interior core of

buildings to allow windows in offices and classrooms.

Throughout all simulations some properties are kept

constant, such as the ASHRAE standard 15 cfm/person of

outdoor air as well as person, lighting, electrical schedules and

power consumption.

Figure 5. HVAC Standard diagram

A zone is susceptible to different loads of varying intensity

and type. In the simulation there are four sensible heat loads:

shell load from the atmosphere, plug load from electrical

devices used inside the room, lighting load is recorded

differently, as lighting can be independent of the number of

people occupying a load. The simulation includes two latent

loads: the atmospheric humidity and humidity from occupants.

The first simulations are done to establish a good base

construction. This includes materials and insulation layering

within walls and roofs. The default constructions are an

inefficient and “heavy wall” design. While the heating load is

lower in winter, the cooling load the rest of the year is greater.

Atlanta is a hot, humid climate, so cooling with

dehumidification is much more significant on energy usage

than heating. The materials are changed to reflect the

construction of the Carbon Neutral Energy Simulation building

on campus that is LEED certified. It was discovered that while

the construction was “lighter” and required more heating in

winter, the cooling load was lowered overall.

Once materials and construction design and schedules were

set, only the methods for heating and cooling the zones were

changed. Like most buildings on campus, the temperature

allowed out of the cooling coil was no higher than 55F, to keep

humidity in the zones acceptable. While this leads to a

comfortable zone, the energy usage is high, due to the amount

of reheating of cold air into the zone. Other systems are

simulated including different versions of heat recovery, radiant

heating, radiant cooling, dedicated outdoor air units,

displacement air, and many others. With the parameters

established, different heating and cooling designs are developed

and tested.

DESIGNS

There are two distinct methods of delivering air into zones:

the variable air volume (VAV) system, as in Figure 6 and

dedicated outdoor air system. Both of these systems were

simulated with heat recovery in the air system or radiant

heating/cooling inside the zones.

Figure 6. Standard outdoor air treatment leading to

variable air volume unit

The Standard outdoor air treatment, as in Figure 7, should

be the most inefficient system because it has to cool all air to

55F before sending it to the supply splitter, where it is then

heated by the VAV (not shown) to insure the room does not

drop below the heating set point.

Figure 7. Standard outdoor air treatment with preconditioning heat and enthalpy exchanger

The standard system with a total heat and enthalpy

exchanger requires balanced flow and affects the outside air


inlet. The total energy heat exchanger includes latent and

sensible heat transfer. Interestingly, it was found to be more

efficient to have a high latent and sensible heat transfer (90%

efficiency) and a low (20% efficiency sensible, 0% efficiency

latent). This is due to the outlet from the refrigeration coils

being set to 55F and there being return air recirculated in the

system. Hotter inlet air requires more cooling energy, so it is

advisable to get the inlet air as close to 55F as possible in order

to accomplish dehumidification.

The dedicated outdoor (DOA) air system is in Figure 8. It

is the simplest, as the intake is balanced with the exhaust.

Figure 8. Dedicated outdoor system

The benefit to the dedicated outdoor air system is that only

the outdoor air needs to be supplied and conditioned for the

space and humidity levels. In addition, the system is zone

independent, which allows different loads and temperature set

points to easily be adjusted for. However, a DOA with heat

recovery, as in Figure 9, is much more efficient, as the air is

precooled and then reheated through dehumidification after the

initial heating and cooling.

Figure 9. DOA with two heat exchangers

The first wheel is sensible heat transfer only while the

wheel on the right is latent transfer only. This has a surprising

benefit because the return air is cooled by giving up its latent

load, which makes the precooling on the left even more drastic.

Conversely, the outside air gets cooled more before being

subject to the cooling coil, but is reheated slightly and

dehumidified more through the desiccant wheel. There is

another benefit, because the inlet air is dehumidified entering

the room, a higher cooling set point can be set because the

dehumidification is not all performed at a single source.

SIMULATIONS The temperature set point schedule, occupancy, lighting,

and plug schedules are all kept constant, as indicated in Figure

10.

time occ

up

ancy

frac

tio

n

ligh

tin

g

load

frac

tio

n

plu

g lo

ad

frac

tio

n

coo

lin

g

setp

oin

t (F

)

he

atin

g

setp

oin

t (F

)

0:00-6:00 0 0.05 0.4 80 65

6:00-8:00 0.5 0.3 0.9 75 70

8:00-12:00 0.8 0.3 0.9 75 70

12:00-13:00 0.5 0.9 0.9 75 70

13:00-17:00 0.95 0.9 0.9 75 70

17:00-18:00 0.8 0.5 0.9 75 70

18:00-24:00 0 0.05 0.4 80 65

weekends 0 5 0.4 80 65 Figure 10. Schedules

Throughout all the simulations, changes were made only to

the system responsible for air flow, heating, and cooling. The

people, plug, and lighting loads were all kept within a typical

load for an academic building to insure reasonable reporting of

results with moderate accuracy. While many simulations were

performed, the most reasonable systems, which covered a broad

range of similar systems that would meet a building or

ASHRAE code, were reported.

A possible source of error is the lack of internally specified

equipment within EnergyPlus. For instance, the fan and pumps

had no default or auto size for power usage; it may be possible

to generate false results with improper fan specification. To

minimize risk in equipment specification, data from high-

performance buildings on the Georgia Tech campus were used

in this simulation, notably Clough Undergraduate Learning

Commons, and Carbon Neutral Energy Solutions. However,

some data, such as those on radiant heating and cooling systems

were underperforming or unavailable. In the event that

specifications were missing or proprietary, a minimal impact

setup was created so as to not set an artificially high load for a

minimal impact system.

Another area of difficulty is the lack of intelligent

simulation adaptation. EnergyPlus requires an extremely

specific inputs to run as intended. This becomes problematic

when sometimes a simulation will run without error, but behave

in a manner that is not representative of the intended model; as

a result, manual checking of the simulation is required to insure

intended and accurate results.

In all buildings, a minimum outdoor airflow is required,

but a minimum of 15 cfm/person was selected in an attempt to

equalize results. However, a traditional VAV does not have

occupancy control and only meets a thermal load and as such,

used the most cooling and heating load.

RESULTS

As Table 2 shows, by reducing the outside air flow in an

identical system, the efficiency of heating and cooling water

systems is slightly offset by the use of electricity. This result

shows that a single answer system cannot be determined. This


becomes increasingly true if there are different methods for

gathering electricity (grid or renewable) or if the chilled and

heated water is produced on site. If an institute can receive

electricity more cheaply than it can supply water throughout a

building, then a high-efficiency VAV system would probably

have cost benefits better than a more efficient chilled beam

system. Comparing raw numbers between buildings is not an

effective method for analysis. To normalize results the use of an

Energy Use Intensity (EUI), both as a sum of total energy (as

kBTU) over gross area and the method of adjusting site/source

usage from the US Environmental Protection Agency [2] was

used in the results. While the EUI of other buildings on Georgia

Tech’s campus is not reported, the expected EUI excluding

source multipliers has been published by the EPA (Autodesk

Sustainability Workshop, 2011) and is in Figure 12.

Figure 11. EPA source energy use conversion

Figure 12. EPA generated EUI table

The setup and usage of passive systems is discovered to

have a drastic change in the efficiency of the building. It is

discovered that having a conventional VAV with chilled beams

saved negligible amount of cooling energy but cost

significantly more electricity. When the radiant temperature

control system is used as the primary method of control with an

occupant controlled outdoor air supply, the electrical, cooling,

and heating load all decreased significantly. It should be noted

that for the DOA and radiant system, an almost constant mass

flow rate of water into the chiller is discovered. While the

details of this anomaly could not be corrected, it is theorized

that by ignoring the chiller power, the DOA with radiant system

is by far the most efficient system. It is believed that a design

criterion within the EnergyPlus software creates a non-zero

mass flow rate through the radiant system.

Table 2. Results

CONCLUSION While the energy usage seems to show large usage with

radiant supplement it is believed this may be due to

inconsistencies within the EnergyPlus model. The dedicated

outdoor air model required a constant mass flow rate of chilled

water regardless of the air flow through the unit, which

dramatically increased chiller power with no gain. Additionally,

in EnergyPlus, heated floors and chilled ceilings cannot exist

within the same floor construction if the zones are stacked.

type

Building

total

elecrical

(MWh)

HVAC

Electrical

(MWh)

Chiller

Electrical

(MWh)

Cooling

thermal

energy

(BTU 10^6)

Heating

thermal

energy

(BTU

10^6)

EUI

(kBTU/ft^2)

EPA EUI

(kBTU/ft^2)

VAV Standard 852 547 285 2630 304 144.3 330.8

VAV Standard w/ Heat Recovery 837 540 305 2467 307 139.0 321.4

VAV People controlled OA 790 485 297 2077 152 121.6 291.9

VAV Radiant heat&cool 1059 754 462 2603 272 160.2 387.3

DOA & Radiant 894 589 305 1202 63 106.6 291.2

DOA & Radiant people controlled 1135 829 694 1068 45 123.1 354.2


Realistically zones on the same side of the building with similar

usage would be calling for heat and cooling at the same time,

meaning a chilled ceiling and heated floor would add to heat

transfer through free convection and further reduce electrical

and hot/cold water requirements. It is also important to consider

the costs of implementing systems when considering where and

how a building is used. In temperate climates where windows

are used, an exclusive radiant system could be acceptable while

in a humid climate dehumidification is important, so radiant

systems can only account for sensible load. What the simulation

shows is how a generic building could perform in any location

with minimal adjustments. A change in the weather file and a

couple of parameters would effectively completely change the

output and the determination of what system is most cost

effective.

REFERENCES [1] "Georgia Tech Stamps Health Services," Georgia Institute

of Technology, 2014. [Online]. Available:

https://health.gatech.edu/Pages/default.aspx.

[2] Trustees of the University of Illinois, Ernest Orlando

Lawrence Berkeley National Laboratory, "EnergyPlus," 2015.

[Online].Available:

https://energyplus.net/sites/all/modules/custom/nrel_custom/pdf

s/pdfs_v8.3.0/EngineeringReference.pdf.

[3] Autodesk Sustainability Workshop, "Autodesk," Autodesk,

2011.[Online].Available:

http://sustainabilityworkshop.autodesk.com/buildings/measurin

g-building-energy-use.

[4] Energystar, "EnergyStar," EnergyStar, [Online]. Available:

http://www.energystar.gov/buildings/facility-owners-and-

managers/existing-buildings/use-portfolio-manager/understand-

metrics/what-energy. [Accessed 25 08 2015].




UNCERTAINTY ANALYSIS METHODOLOGY FOR PARTICLE HEATING RECEIVER TESTING

Clayton Nguyen Georgia Institute of Technology Atlanta, Georgia, United States

Matthew Golob Georgia Institute of Technology Atlanta, Georgia, United States

Sheldon Jeter Georgia Institute of Technology Atlanta, Georgia, United States

Said Abdel-Khalik Georgia Institute of Technology Atlanta, Georgia, United States

ABSTRACT The use of particulates as a thermal medium for solar

receiver technology has opened new possibilities in pushing temperature limits and power cycle efficiencies for solar power plants. To explore particulate use, a new type of solar receiver is being developed at Georgia Tech. To show the viability of this technology a high thermal efficiency in combination with highly accurate results must be shown.

In this paper the uncertainty analysis used to evaluate solar particle heating receivers will be covered. The paper focuses on the uncertainty of single point measurements and on reducing the direct measurements to obtain an overall lower indirect measurement uncertainty. The end result is that a receiver efficiency with an uncertainty lower than 5% can be found.

INTRODUCTION This paper will describe and demonstrate the methodology

and data sources used to analyze and estimate the combined uncertainty of the experimentally measured solar energy collection efficiency of a small scale particle heating receiver. While such a value can be calculated through basic thermodynamic and statistical techniques, the greater objective of this work is determine the receiver efficiency with uncertainty that is less than 5%. Uncertainty results will come from a combination of two types of error, random fluctuations (UA) and systematic bias (UB). A thorough accounting of the methods used to obtain an uncertainty of less than 5% will be shown here. These methods are recorded to show the validity of the results of the experiment and to provide a guideline and reference for future related work. The experiment uses Georgia Tech’s High Flux Solar Simulator to heat a particle heating receiver (PHR). To test this apparatus three main measurement methods are used. A calorimeter is used to measure the expected heat flux into the particle heating receiver, a scale base is used to measure the outlet mass flow rate and thermocouples are used to measure the heating of the particulate through the PHR.

LITERATURE Presently, we are collecting single point efficiency data and

not developing a statistical model for the efficiency. The basic theory for the uncertainty of single point measurements is based on the work of Kline and McKlintock [1], describing how to compute the uncertainty for single sample experiments. The basic form for calculating an indirectly measured value is shown in equation 1 below.

22

22

2

11

nn

RB wv

Rw

v

Rw

v

RwU

(1) The paper describes how to combine the uncertainty of the

various variables into an overall uncertainty for the final computed measurement. For completely independent sources of error the sum of the squares of the uncertainties is equal to the squared overall uncertainty, as in equation 2.

222BAC UUU (2)

Standards set forth by Kline and McKlintock are used to help determine the American Society of Mechanical Engineering’s standards for uncertainty analysis [1]. The measurement of solar collection efficiency is particularly complex in comparison with similar mechanical engineering applications, so special attention is required to attend to all the details.

EXPIREMENTAL BACKGROUND The purpose of the experiment is to find the receiver

efficiency of a PHR. To do this HFSS is used to heat the particulates that flow through the PHR. In the current experiment particulates flow through the receiver by process of a batch system. An upper hopper is opened to allow the particulate to fall through the system, and the particulate is then caught in the lower hopper.

The receiver efficiency is evaluated using equation 3:


rcalorimetercalorimete

inout

rCalorimete

absorbedPHR

)(

Q

dTcm

Q

hhm

Q

Q

out

in

T

T

p

(3) To evaluate the receiver the efficiency, the upstream

temperature, downstream temperature, specific heat of the particulate, irradiated heat flux in and mass flow rate must all be measured. To minimize the overall uncertainty each of these measurements is carefully calibrated: Mass and mass flow

The lower hopper is placed on top of a load cell; the response can be used to transiently measure the mass flow rate of the system. The calibration of the load cell is the first of the key measurements necessary for evaluating the PHR. The change in the signal output of the load cell over time will be used to measure the mass flow rate.

Incident power

The second key measurement is the irradiative power going into the PHR, which is measured by a separate device, the calorimeter. The calorimeter uses the same receiver aperture as the PHR, in the same location to ensure similar operating conditions.

Temperatures

Finally to measure the particulate temperatures, a series of thermocouples is placed within the system. Each of these thermocouples must be submerged by the particulate flow to read an accurate temperature. They must then be calibrated to a higher degree of accuracy due to the small expected temperature differential. Specific heat

The specific heat of water is a well-known value. Water purity is measured to assure that there are no significant deviations from the commonly accepted values. These values are used when calculating the heat input into the calorimeter.

In comparison, the specific heat of the particulates is much more difficult to measure but serves an important role since these tests are to be conducted anywhere between 25°C to 700°C. Over this range of temperatures the specific heat changes significantly. To accommodate for this, the specific heat values will be found using experimental and chemical composition data.

STATISTICAL UNCERTAINTY While the bulk of this paper focuses on the minimizing the

systematic bias in the results, each set of tests and calibrations will also take into account the statistical uncertainty. The statistical uncertainty is obtained by calculating the sample standard deviation of a set of data, multiplying that by the corresponding coverage determined from the small sample t-statistics factor and dividing the square root of the number of data points. UA is the expanded uncertainty with 95%

coverage, while the UA is the standard uncertainty, as in equation 4.

n

SSDkukU cAcA

(4) The coverage factor, kc can be found by taking the two

tailed inverse of the Student’s t-test at a 95% probability. For estimating purposes 2 can be typically used as the coverage factor for large data sets. At the end of the experiments the UA is found over a steady state result (explain s-s) and is then combined with the UB to find the combined uncertainty, UC.

TRANSIENT MASS FLOW MEASUREMENT The measurement of the mass flow rate of the systems is

absolutely essential to transiently measuring the receiver efficiency of the system. A scale base [2], load cell with a platform to prevent oscillation issues, is used to measure the mass of the system. Post analysis of the signal allows for the mass to be measured with time to find the mass flow rate.

The scale base is calibrated using two different methods. The first method uses a series of weights measured by a high accuracy lab scale (±0.1 g). This provides a calibration range for up to 50 kg. This calibration is used for measuring the mass of a gradually filling bucket of water in order to calibrate a flow meter. The scale base provides an additional frame, which is used to mitigate the slight oscillations that occur in isolated load cells. This in turn reduces the noise in the signal output from the load cell.

Mass calibration of the scale base used to measure the mass flux rate was done incrementally using an existing hopper filled with ID50-K particulate and an empty hopper set on a scale base mounted in the planned PHR test configuration. A beaker full of ID50-K was then taken out of the full hopper placed on an OHAUS GT8000 (±0.1g) scale [3], where it was weighed and recorded with the beaker’s mass tared out; it was then poured into the empty hopper. This process was repeated until the empty hopper on the scale base was filled. A few extra weights were added to increase the calibration range in case of future modifications to the hopper. This yielded a calibration for the scale base accurate to within ±2.5g and ±0.21g/s transiently over the needed test range.

The GTHFSS setup uses the manually calibrated scale base load cell in order to measure the mass in the outlet hopper. Post processing was used in order to transiently measure the mass flow and flux rates. The rate of change in the load cell signal was calculated using a five point stencil as in equation 5 below,

avg12218182

tiiii

dtid

(5) where tavg represents the average time differential between each measurement, ε is the signal voltage and i is the index for each measurement. A calibration constant Km was then used to convert the data into a mass flow rate as in equation 6.


CalKdt

dm

(6)

Table 1: Expanded Uncertainty for the Particulate Mass Flow Rate

Measurement xiU

Influence Coefficient,

ix

m

22

ixii x

hUU

(g/s)2

2

2

B

i

U

U

(%)

Source

ROC of Scale Voltage, ε

1.026 x10-8 V/s 71002.2

)/(

mK

dtd

m

9.8×10-2 100 (1)

Scale Calibration Constant, K

1640 g/V

6105.6

dt

dmK

m

1.1×10-4 0 (2)

sum of Ui2 = 9.8×102 Expanded

Uncertainty UB= 3.1×10-4 kg/s

1) 5 point numerical stencil, Uxi calculated in Appendix 2 using the 34970 user manual (1)

2) Manual calibration using a series of known weights

Figure 1: Mass Calibration

By taking a graph of the accumulated mass against the time one can see that the average mass flow rate is 0.13 kg/s with an R2 value well above 0.90.indicating a steady mass flow rate.

CALORIMETER EXPERIMENT The calorimeter is the simplest of the devices and

quantifies the heat input into the system through the HFSS. The calorimeter is made up of a copper tube coiled into two sections, a cylinder and a cone. The coils of the copper tube are then compressed using a stainless steel frame, and a cap is added to end of the cone in order to form a receiver. The cylindrical diameter of the calorimeter is 102 mm, which is larger than the PHR aperture. The larger diameter of the cylinder is to ensure that the entirety of the incident radiation is captured within the cavity.

Figure 2: Calorimeter Overview (1) Insulation, (2) Water

cooled coil, (3) Cavity, (4) Stainless Steel frame Water flows within the copper tubes to absorb the energy

absorbed by the cavity. Water temperatures are measured at the inlet and outlet of the calorimeter using resistance temperature detectors (RTD). The calculated enthalpies are then used to find the amount of heat input expected in the PHR experiment. Each of these RTDs is calibrated using a highly accurate standard from Burns Engineering. The platinum resistance thermometer (PRT) has an accuracy of ±0.0025 K and can be calibrated anywhere between -190°C to 420°C. The PRTs are calibrated using a water bath at several different set points, and this is done during heating and cooling to prevent uneven heating and cooling biases.

Additionally a digital flow meter measures the mass flow rate of the water. The device provides a voltage response based on the flow of the water. The mass flow rate is obtained as an indirect measurement. This is accomplished by calibrating the flow meter while filling a container placed on the previously calibrated scale base. To provide an added sanity check the final mass and time are compared to the mass flow rate for that time period.

Measurements are taken at three different flow rates. A five gallon bucket is used to hold the water; as such the maximum amount of statistical data that can be collected for each run is dependent on the time necessary to fill the container. Each of the runs is conducted using 100 powerline cycles, and a measurement is taken every 10 seconds. At the highest flow rate 40 measurements can be taken prior to overflow.

These measurements are taken using a digital instrument and have the associated resolutions for each device. The calibration of each of these devices determines the least significant digit. The uncertainty of the calibration is taken to be the systematic bias in each single measurement experiment.

y = 0.13x + 27R² = 1

0

20

40

60

80

0 100 200 300

Mass Accumulated (kg)

Time (s)

1

3

2

4

UAB

CALOSIMU

Tsolar precisappar1” (2geomto a measuthan 4underwith accurwith l

APET

caloriapertuthe twthroubetweplate confirFigurplate

Figu

P

readintargetmaketest stpositi

School of Eng

ORIMETER ULATOR LAMThe current testflux on the re

se determinatioratus (calorime25.4mm) behin

metric configuraknown area. ured at the GT4%, a thresholr. As such an the PHR’s ap

rate heat inputless than 4% re

RTURE SETUTo ensure thaimeter test wiure (iris) platewo devices. T

ugh which theeen the receive

will remain rmed through re 3. The caloriin order to me

ure 3: Laser p(2) Iris

Prior to placemngs were also t, Figure 3. O

e a reference pltand used for bioned in the G

2

gineering – Me

MEASURMPS

t setup is desigeceiver which on of the lab seter) is positionnd the aperturation that constIn addition th

THFSS in phasld which our c

updated caloperture plate t, Qin, to attaielative uncertai

UP at the solar fll be similar t

e in the same lThe iris plate incoming areer and calorimthe same bethe use of a s

imeter will be asure the heat

positioning dets Plate, (3) La

ment of the caltaken at the fo

Once completedlane 25.4mm aboth the PHR aTHFSS. The te

1

echanical Engin

EMENT O

gned to accuratis a critical co

scale PHR effined such that thre (iris) plate,trains the incom

he prior PHR e 2 had an unccurrent experimorimeter used will be used in an overall Pinty.

flux that occuto the receiverlocation will bprovides a coea can be asseter tests. The tween the twset of planar laplaced directlyinput.

tails (1) Focaambertian Shie

lorimeter, fluxocal plane usind a planar lasway from the fand calorimeteest stand also h

3

neering - ECT

OF SOLAR

tely quantify thomponent in thciency. The tehe focal plane which is in ming lamp lighefficiency dat

certainty greatement aims to bin combinatioto measure a

PHR efficienc

urs during thr test, the same used betwee

onstant apertursumed constanposition of th

wo tests and asers, shown iy behind the ir

l Plane Lasereld

x maps and flung a Lambertiaser was used tfocal plane. Ther tests was thehas a mount fo

TC 2015 Procee

R

he he st is a

ht ta er be on an cy

he me en re nt is is in is

r,

ux an to he en or

the Lathe latestingusing purposhieldverifyimageand thcalorim

FiguFlux

In

criticacriticairis, cplane.calorim0.001m

HEATTo

calorimprobesreceivlossesfeed aalso pflow riris plreceiv

Ttypicaas in e

edings – Vol. 1

ambertian targeamps reach stg. The location

the establishses, flux gaug

d locations imy continuity ines were taken bhe GTHFSS symetry tests

ure 4: (Left) Lx image of lam

n most aspects al factor. To mal in accuratelycalorimeter, an. This ensured meter measurem.

T FLUX MEATo measure themeter’s coppes located on th

ver. The whos from the copand discharge lplaced upstreamrate of the watelate in place,

ver can be estabThe heat flux thal calorimeter fequation 7.

mQ

14

et to act as a steady state bn of the shield hed planar l

ge data and flummediately bef

n lamp perforbefore any PHRystem were in

Lambertian tarmp 7 at focal p

of testing the manage this, thy gauging and nd Lambertian

d a consistent aements when

SUREMENT e lamps’ outpuer coils with he water inlet aole body is hepper non-cavitlines. A displam of the calorer passing throu

a heat flux oblished. hrough the irisformula divide

waterp,watercm

1

shield. This isefore calorimwas confirme

lasers. For qux images werfore each calormance. The sR runs to verifysame conditio

rget at focal pplane Flux Ga

calorimeter, lohe series of pla

setting the orin target to thand quickly retesting, with a

ut, water was rcalibrated RT

and outlet of theavily insulatety surfaces as

acement water rimeter to meugh the calorimoutput of the

s plate is calcued by the area

2

4inoutr

DTT

Page 13

used to ensurmeter and PHR

d between runquality controre also taken aorimeter test tosame data any that the lamp

on as during th

plane (Right) auge Sensor

ocation was thanar lasers waientation of the lamps’ foca

epeatable set oan accuracy o

run through thTD temperaturhe coiled cavityed to minimiz well as wateflow meter waasure the masmeter. With th

lamp into th

ulated using thof the aperture

2 (7

1

3

e R ns ol at o d s e

e as e

al of of

e e y e

er as s e e

e e,

7)


Table 2: Uncertainty Table for 4 Lamps Calorimeter Test (~123.5 W)

Measurement xiU

Influence Coefficient

ix

m

22

ixii x

hUU

(W/cm2)2 2

2

B

i

U

U

(%)

Mass Flow Rate1 ṁ

0.22 g/s

7.22

inoutwaterp,4

)(

D

TTcwaterm

Q

0.35 21

Inlet Temperature2

Km

0.0251 °C

6.32

waterp,water4in

D

cm

T

Q

0.0081 0.50

Outlet Temperature2

Tout

0.0251 °C

6.32

waterp,water4out

D

cm

T

Q

0.0081 0.50

Diameter3

D 0.038 cm

293

inoutwaterp,water8

)(

D

TTcm

D

fluxQ

1.3 78

sum of Ui2 = 1.6 Expanded

Uncertainty UB = 1.3 (W/cm2)

1) Mass flow rate is obtained from flow meter calibration, which is performed using a scale base that has been previously calibrated using known weights.

2) Temperature readings are calibrated using a Standard Platinum Resistance Thermometer

3) The diameter is found using multiple measurements taken using a dial caliper

Table 3: Heat Flux for Various Lamp Combinations

Lamps Overall Heat Rate

(W) Heat Flux (W/cm2)

1, 4 2760 ± 15 55 ± 0.58 1,4,7 4470 ± 23 89 ± 0.93 1, 2, 4, 7 6030 ± 31 120 ± 1.3

In order to determine the effects of heat loss in the calorimeter, the calorimeter was run at ambient temperatures. The increase in the water temperatures and the flow rate was used to calculate the UAL (overall heat loss conductance) as shown in equation 8 below.

2outletinlet

rCalorimete

)rCalorimeteamb

(L

inoutwaterp,waterL

TTT

TTUA

TTcmQ

(8)

This UAL was then used to calculate the heat loss from the calorimeter for tests at non ambient temperatures. The temperature of the calorimeter was approximated by taking the average of the inlet and outlet temperatures of the water.

Even with the highest heat input used, 6200W, the overall heat loss was calculated to be only be 9.1 Watts. Relative to the input this is negligible at lower temperatures. Fig. 5, below shows the heat flux once the lamps have reached steady state operation.

Figure 5: Lamp Solar Flux Outputs

It is important to note that the number of lamps used on the

calorimeter is only approximately proportional to the solar flux through the aperture. This is due to complexities in the system causing each lamp to provide a different amount of power. In addition, the amount of irradiation that is focused through the aperture differs from lamp to lamp due to slight variations in the alignment of the lamps and interference from the iris edge.

RECEIVER TESTING The particle heating receiver is tested using a batch process

system. Particles are contained within an upper hopper that provides enough material for approximately 5 minutes of run time. The scale base is used to transiently measure the mass flow rate of the particulates through the particle receiver. To measure the temperature of the particulate, sets of thermocouples were placed at the inlet and outlet of the particle heating receiver.

The outlet particle thermocouples measure the particulate temperature after the particles travel through a mixer section. Due to the low thermal conductivity of the particles and the uneven heating in the PHR at the small scale, this section is required in order to measure any sort of average particle outlet temperature.

The receiver efficiency is calculated by comparing the difference of the inlet and outlet enthalpies to the heat input measured by the calorimeter, as in equation 9. These enthalpies are highly dependent on the specific heat of the particles, which

0

50

100

150

0 20 40 60 80So

lar Flux (W

/cm^2

)Scan

Lamps 1,2,4,7 Lamps 2,4,7 Lamps 1,4


is temperature dependent. To measure this specific heat, a digital scanning calorimeter (DSC) can be used in combination with a theoretical model using the Kopp-Neumann law. According to the manufacturer, a high end DSC, assuming no operation errors, can provide an accuracy of 3.5%. In comparison, studies have shown that the Kopp-Neumann law is accurate to about 3% near ambient and 4-6% at high temperatures.

rcalorimetercalorimete

inout

rCalorimete

absorbedPHR

)(

Q

dTcm

Q

hhm

Q

Q

out

in

T

T

p

(9)

To evaluate uncertainties related to the Kopp-Neumann law, a conservative uncertainty will be assumed. For DSC measurements, 3.5% uncertainty is used as the UB while the UA is calculated during the regression of an experimental model for the data.

The uncertainty table used for calculating the receiver efficiency shown below is an abbreviated table. Since there are multiple thermocouples, and since the numbers change from setup to setup, the table easily changes depending on the setup of the instrumentation for the PHR. In addition, depending on the type of numerical regression used, the model used for the formula changes drastically and can greatly extend the size of the partial derivative. As such this table serves only as a basic template.

Table 4: Uncertainty Table for Receiver Efficiency Test

Measurement xiU

Influence Coefficient, `

ix

m

2

2

B

i

U

U

(%)

Mass Flow Rate, ṁ

0.0031 kg/s

476.9PHR

m

45 (1)

Inlet Temperature, Tin, 1

0.025 °C 02.0PHR

inT

0.01 (2)

Outlet Temperature, Tout, 1

0.025 °C 01.0PHR

outT

0.01 (2)

Calorimeter Heat Input, Qcalorimeter

0.001 kW 14.

rcalorimete

PHR

Q

0.5 (3)

Specific Heat Cp

3.5% 92.0PHR

pc

54 (4)

1) Obtained from flow meter calibration, which is performed using a scale base that has been previously calibrated using known weights.

2) Calibrated using a Standard Platinum Resistance Thermometer

3) Measured using a digital scanning calorimeter and the Kopp-Neumann Law.

4) Experimental measurement

DISCUSSION The receiver efficiency for an ambient test run was found

to be 91.58% with a UA of 0.61% and a UB of 4.36%.The major sources of error can be attributed to uncertainty in the specific heat and the mass flow rate. The uncertainty in the specific heat was not unexpected due to the inherently difficult task of measuring the specific heat of particulates in comparison to measuring that of a block of material. These difficulties can be attributed to the interstitial spaces between the particles and to difficulties due to contact resistance.

CONCLUSION This methodology has used several different instruments to

measure data in order to evaluate the receiver efficiency for a particle heating receiver. Each instrument has been calibrated as accurately as possible to minimize uncertainty in the results. As such the main sources of error come from the uncertainty of the specific heat, which cannot realistically be reduced further than 3.5%, and the mass flow rate. With an absolute uncertainty of about 4.5% this methodology for measuring the efficiency of particle heating receivers provides a means by which future experiments can be expected to provide high quality results.

REFERENCES 1. Describing Uncertainties in Single-Sample Experiments. Kline, S. J. and McClintock, F. A. 1, 1953, Vol. 75. 2. Ohaus B250S. Champ Scale Bases. [Online] dmx.ohaus.com/WorkArea/downloadasset.aspx?id=6570. 3. GT8000. Ohaus Discontinued Product Instruction Manuals. [Online] [Cited: ] 4. Keysight Technologies. 34970A Data Acquisition / Data Logger Switch Unit. Keysight Technologies. [Online] May 2012. http://www.me.umn.edu/courses/me4331/FILES/AgilentManual_34.pdf.




A DISTRIBUTION LEVEL RENEWABLE INTEGRATION OPTIMIZATION TECHNIQUE WITH VARIABLE PRICING, DEMAND RESPONSE AND THERMAL ENERGY

STORAGE FOR RESIDENTIAL APPLICATIONS

Justin M. Hill Hessam Taherian Sandeep S. Chahal

University of Alabama at Birmingham Birmingham, AL, USA

ABSTRACT

As renewable energy generation continues to proliferate

throughout the world, advanced strategies for integrating

intermittent generation into the existing grid infrastructure must

be developed. This paper proposes a technique where an

iterative method of energy supply and demand optimization in a

distribution system can be implemented by utilizing

advancements in two-way communications, variable rate

structures, energy storage and home energy management

(HEM) systems. This approach coordinates traditional

resources with renewable generation and energy demand to

maximize the production from renewable resources and

minimize the curtailment due to excess supply. It begins with

the electric utility developing a sub-hourly day-ahead energy

cost based on forecasted renewable output and energy demand,

which is then sent directly to a participating customer’s HEM.

This pricing structure contains multiple energy price points

each hour, typically in five to fifteen minute intervals, which

are based on the forecasted energy availability for the

upcoming day. The customer’s HEM takes this cost schedule

and develops a schedule for how the home will operate based

on programmed preferences and forecasted distributed

generation production locally available, all with the goal of

minimizing the total energy cost for the homeowner while

maintaining their comfort settings. Each home then sends their

load shape to the utility who aggregates the energy

consumption data and recalculates an updated sub-hourly

pricing scheme. This new pricing schedule is republished to

further encourage customers to move what energy consumption

they can to lower cost times where energy generation is in

excess of demand. This process will iterate until the supply and

demand of energy are reasonably aligned, utilizing additional

necessary parameters required to ensure the optimal strategy is

seen by all participants. Additionally, customer owned energy

storage systems – thermal, battery or other – receive a second,

more real-time pricing signal throughout the actual day to

compensate for forecasting errors in the day-ahead scheme.

The goal of this strategy is to develop a non-invasive method of

coordinating energy supply and demand while allowing the

customer to maintain control over their energy usage and

comfort. The strategy is currently being modeled with

approximately ten homes utilizing DOE building simulation

platforms and tools which can show the effect of the control

algorithms when aggregated together in a community.

INTRODUCTION Since energy generation is beginning to shift away from

largely base load sources with fossil fuel and hydro based assets

being utilized to compensate for the variability in energy

demand towards higher amounts of renewable generation,

which introduces uncertainty in both supply and demand of

energy, energy utilization optimization research will be forced

to follow. This implies a shift in research from component

level efficiencies to controlling when loads occur on the grid

and making them responsive to grid signals. More simply

stated, it will become increasingly important at what time of the

day energy is consumed rather than how much energy each

component consumes over the course of the year – see [1].

This type of responsiveness in end-devices to grid signals is

made possible by the advent of sophisticated communication

protocols and/or advanced, second generation smart meters

from electric utilities which can be used to exchange pricing

signals and other types of information and also, more

importantly, the advances in wireless communications and the

increase in their speed and bandwidth make fast exchanges of

large packets of data possible.

Due to the previous factors, the project described in this

report focuses on a control methodology to iteratively optimize

the balance between supply and demand for energy on a day-

ahead basis utilizing variable rate structures. The project also

implements a secondary rate structure to utilize different types

of energy storage on a shorter time interval. This dual

timeframe structure was chosen, as it is seen as a void in

current research and can be utilized by the utility to allocate

generation resources more effectively and increase the

efficiency of the grid while also compensating for errors in the

renewable energy output and demand side forecasts. This


approach is also seen as a way to minimize inconvenience or

impacts that occur from real-time control of devices or other

type of direct load control through traditional forms of demand

response.

There have been several research projects found related to

this topic and serve as a foundation for this project; however

the available research focuses on different objectives. These

include factors like solely minimizing the cost to the customer

and do not allow the utility to utilize the information to

optimize the grid and benefit all consumers served [2], [3], [4],

[5], or they focus on only one technology in the home [6], or

plan to operate in a continuous real-time manner [7], [8], [9],

[10] requiring unnecessarily high levels of computing power,

communication bandwidth, with a high potential to cause

customer inconvenience.

Previous work in this field has been performed, most

notably the work done by Bakker et al. in [7] and [11] which

presents a three step optimization methodology which includes

building a daily load shape for the house and iterating with the

utility for an optimal solution but makes operating decision for

all appliances in real time using a centralized controller. This

approach stops short of utilizing any inherent thermal energy

storage in the home and its appliances. Another noteworthy

project is presented by Li et al. in [2] which presents a method

to determine the daily energy usage of a home and optimize its

performance based on a learned thermal model of the home

from thermostat data. The research of this paper builds upon the

separate research performed by these two entities but also

includes major differences such as a) time horizons for

scheduling of different appliances, b) overall simulation goals

of incorporating renewable generation sources using demand

side resources and energy storage to supplement the grid rather

than having an additional local fuel based generation source c)

optimizing the grid rather than solely minimizing the energy

costs to the customer and d) utilizing the same pricing signal to

all customers at all times rather than steering customers

individually with customized cost signals. These four major

differences are seen as gaps in the present research in the area

and can be used to improve the work done previously while

also making it more relevant to current US energy market.

The first change (a) allows for sub-hourly planning of

individual appliances through the use of pricing signals,

increasing the granularity of accuracy while also adding rules to

the algorithm to prevent the systems from short cycling by

implementing a minimum run time. This also includes a more

real-time approach to send a second pricing scheme to energy

storage devices which can be used to compensate for errors not

seen the day before. The second change (b) increases the

complexity of the algorithm, since there is no longer a local

fuel source to compensate for errors; however the addition of

energy storage, both thermal and electrical, can be used. Also

by removing the local generation source, it aligns more closely

to the energy market in the US where large, central generation

plants are dominant and can allow for this type of system to be

provided as a service to the customer and the utility without

additional generation. The third difference (c) simply shifts the

overall focus of the research from a self-serving algorithm to

minimize the cost for one individual to a system which allows

the players involved to minimize their energy costs while also

working together to optimize the grid and decrease the amount

of fossil fuel based generation required to meet their needs.

Finally the fourth change (d) is done to maintain fairness to all

customers involved. This change increases the complexity of

the algorithms required but is seen as the only feasible option

for field implementation due to the amount of government

regulation in the utility industry and maintaining an unbiased

control algorithm.

The paper is organized as follows. First we introduce the

overall control strategy and discuss the merits and limitations of

it. Secondly, the modeling approach is discussed and broken

into work that has been completed and modeling that is still

planned. Finally we discuss future research plans for the

project.

CONTROL STRATEGY The proposed control algorithm is built using six major

steps.

1) The utility builds a day-ahead, sub-hourly pricing

schedule based on expected demand and energy

generation. This is very similar to how the grid

operates today and just assumes an increased level of

renewables.

2) It sends this pricing schedule to the customer’s HEM.

3) The HEM develops a schedule of appliance operation

based on historical customer usage, customer

preferences and cost of energy.

4) The customer’s HEM sends their sub-hourly load

shape back to the utility.

5) The utility aggregates this load shape information,

combines it with utility owned energy storage and

republishes updated sub-hourly energy costs.

6) Steps 1-5 are repeated iteratively until an acceptable

alignment of supply and demand are found.

These six steps give the utility and the customer a baseline

load shape, and the customer’s system decides when appliances

will operate based on historical data. However there are three

main concerns that come from this setup, the first is having

customer’s shifting all their energy usage to one time frame

where the lowest cost period is seen, thus making optimization

impossible. This can be avoided by adding a cap to the amount

of energy available at the lowest cost by introducing an

inclining block rate as described in [12]. This encourages

consumers – and their HEM – to spread out their energy

consumption to more evenly match the supply of generation to

the system. The second concern is compensating for errors in

both supply and demand forecasting. To compensate for this,

the control strategy will provide a separate, near real-time price

that is provided to energy storage devices, both thermal and

electrical e.g. water heaters, electric vehicles and battery energy

storage systems. This type of dual-pricing policy was

demonstrated in [13], which sets pricing based on different

electrical tasks – interruptible or non-interruptible and is


utilized to flatten a customer’s load shape. This differs from

our proposed control strategy, as the proposed strategy only

targets energy storage devices rather than all appliances in a

home and also has the goal of compensating for forecasting

error rather than simply flattening a customer’s load shape,

making it a more flexible platform. The third issue is a scenario

where an HEM would attempt to manipulate the system by

sending inaccurate data to the utility and force the energy price

down during hours of the day that only benefit them rather than

the system as a whole. This is addressed with the inclining

block rate to a certain extent, since a large consumption in a

period of time would cause planned usage over that amount and

increase the cost of energy for that customer. However to

discourage a system from projecting energy usage substantially

higher than is actually going to be used, a penalty for under

usage would then need to be enforced – this would require a

tolerance level be developed as to not punish well-meaning

customers. This will become a moot point after the system

reaches a large enough scale, because each customer is not

privy to other customer’s load shapes or information, so a large

over-bid from a single home will not impact the price and

therefore not be in the best interest of the customer to attempt

to mislead the system.

Overall, this method of control is chosen because it is

applicable to all types of systems on a neighborhood level

scale. That is, it can be applied to a traditional neighborhood

where there is no local renewable energy or storage or in a

location that has centralized or distributed and/or community

scale renewables and energy storage. The system is also

designed to optimize the grid with very little interaction from

customers themselves and each system learns over time to

match the customer’s preference in scheduling and balancing

their needs of cost with comfort and convenience. It should be

noted that the intent of this control strategy is not to be the only

solution for integrating renewable energy and is not designed to

handle fast response items such as frequency regulation. The

purpose of this strategy is to develop a low cost method to

optimize how the grid operates with increased renewable

generation and to provide customers the ability to take

advantage of this to lower their energy costs while also

benefiting the grid.

MODELING THE SYSTEM The control strategy described above must be modeled to

determine gaps and its effectiveness as a resource for

integrating renewable energy into the grid more effectively.

This section includes modeling that has been performed as well

as modeling that will be taking place over the next several

months. The future modeling section is based on a plan from

the knowledge that has been developed to this point, however

results from earlier modeling sections can influence and change

decisions as necessary.

The first step in modeling the system is to develop a

building energy model of a home and verify that it is thermally

accurate and representative of how a home would respond with

the control algorithms in place. To do so, multiple weeks of

circuit level sub-metered data and thermostat setpoints were

gathered from a home in the Birmingham, AL area where the

physical characteristics and appliance types are known. This

information was then used to develop schedules for appliances

such as the refrigerator, washer, dryer, dishwasher, etc. and also

used as a baseline to determine how well the model is

calibrated to the thermal characteristics of the home and the

simulated HVAC usage was compared to the metered data. The

home was initially modeled using BEopt 2.3 [14], a front-end

to the DOE’s EnergyPlus building energy simulation tool [15]

which was developed to assist engineers with introducing

energy efficiency upgrades to homes. However, since sub-

hourly schedules and other information was available, the

EnergyPlus text editor was also used to input more detailed

information and schedules to mimic the energy consumption

patterns seen in the metered data. A 3D sketch from BEopt 2.3

of the home is shown in Figure 1.

Figure 1. 3D Sketch of Model Home.

In addition to the home characteristics, to produce a

realistic model of the house, a weather file containing actual

measured data for the time period of the model is required. To

create this, weather data was gathered from the Iowa

Environmental Mesonet [16] website which includes historical

weather data for locations across the country. A second source,

SolarAnywhere [17], was also used to include solar radiation

data which was not present in the original weather dataset.

After the envelope, energy consumption of appliances and

weather data was developed, simulations were performed to

match the HVAC energy consumption and the whole home

consumption between the metered data and the simulation over

the time period available – August 5, 2015 – August 21, 2015

by changing unknowns in the makeup of the home such as

infiltration rates. After some minor envelope changes the

simulated and the metered load shapes began to line up. This

data is shown in Figure 2 and contains the simulated and actual

energy usage data measured for both the HVAC (top) and the

whole home energy consumption (bottom). As can be seen, the

lines do not match exactly as there are several small

introductions of error including the meters used for sub-

metering, minor physical characteristic differences of the home

and the model and local weather variations from the monitored

site.


Figure 2. HVAC (Top) and Whole Home (Bottom) Usage, Measured vs. Simulated

Although the data does not align exactly from the model to

the metered data, it is seen to be very representative of the

home’s thermal characteristics. Data shown in Table 1

summarizes the differences in the total energy usage and peak

demand of both the HVAC system and the whole home. This

data shows a minimal difference between the simulated and the

metered data, demonstrating accuracy in the thermal

characteristics of the model.

Table 1. Comparison of Metered and Simulated Usage

Usage

(kWh)

Peak Hourly

Demand (kW) Percent Diff.

Metered HVAC 305 3.5 4.2%, 2.9%

Simulated HVAC 318 3.4

Metered Home 470 6.8 1.7%, 14.2%

Simulated Home 478 5.9

Once the model has been validated to be thermally

representative of the home chosen, the next step to model the

project focuses on developing a starting point for a real-time

pricing (RTP) rate. This information will ultimately be utilized

as the day-ahead rate structure initially sent to the included

homes in the model. To compile this information, historical

hourly capacity energy price data from the ERCOT market [18]

was gathered to provide a proxy to correlate the energy demand

on the grid for each hour of the year.

The most recent year available for the ERCOT capacity

pricing data was 2014 and this data will serve as the foundation

for which RTP rates are chosen, however another important

factor in their determination is the renewable energy output. To

include this variability in the pricing model, five minute

interval data of simulated energy generation for both solar PV

and wind generation was gathered from the National

Renewable Energy Lab’s (NREL) study on the eastern

transmission renewable integration study [19] and their Wind

Prospector tool [20]. Two sites for wind (16 MW capacity

each) and two sites for solar PV (39 MW capacity each)

generation were chosen in the Birmingham, AL area. The most

recent data available for wind energy was from 2012 and that

for solar PV was from 2006. While it should be noted that none

of these three date ranges align, it was determined to be

insignificant since the modeling is based on typical weather

data and a typical year of performance rather than validating

against a known baseline.

A typical daily load shape for solar is relatively constant,

with the output rising as the sun comes up, peaking at around

noon and then decreasing output as the sun goes down. Wind

output is less predictable but tends to follow a consistent output

when averaged over longer periods of time. However, the

variations throughout the day are more important for this

research and can cause the energy price to vary from day-to-day

and potentially even minute-by-minute. This variability can be

seen in Figure 3 and shows the average percent output

compared to its maximum generating capacity over a fifteen

minute period.

Figure 3. Fifteen Minute Energy Output (%) [19], [20]

While understanding this variability exists with renewable

generation sources, for this project we can combine the four (or

however many sites were to actually exist) into one load shape

since we focus on how a distribution level system can respond

to the grid fluctuations rather than matching the output of an

individual renewable generation source. Figure 4 shows the

same day combined weighted output from the four generation

sources from Figure 3.

Figure 4. Combined Renewable Energy Output –Weighted [19], [20]

This data shows a smaller peak output of just over 50%

compared to almost 70% before but shows some smoothing of

the load shape, making predicting more feasible but still not a

highly accurate process. This introduction of error makes

electrical energy and thermal energy storage even more

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0:00 3:36 7:12 10:48 14:24 18:00 21:36

Per

cent

Outp

ut

(%)

Wind 1 Wind 2 Solar 1 Solar 2

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0:00 3:36 7:12 10:48 14:24 18:00 21:36

Per

cent

Outp

ut

(%)


valuable and is the purpose of the proposed dual-pricing policy

to have these devices compensate for the inevitable error in

predicting output.

This renewable generation information must now be

incorporated into the pricing scheme described previously. To

do this, a correlation must be made between the percent output

and the cost increaseor decrease it causes. This was done by

creating a multiplier at each time step to be applied to the

baseline energy cost for the renewable energy. Since the

correlation will not be linear from zero to 100%, four evenly

distributed segments were chosen with different slopes and a

cross-over point at 50% output. This correlation with the

respective formulas can be seen in Figure 5.

Figure 5. Renewable Energy and Cost Curve

Correlation To combine the two sources of cost it now becomes a

simple algebraic equation where the ERCOT data is used for

the traditional generation load (67%), and the cost data found

from Figure 5 is used to represent the renewable generation

output (33%). These percentages were chosen to align with

California ‘s renewable portfolio standards [21] and are a

simplification of the breakdown of cost allocations. Other

variables necessary for this calculation are the cost to generate

renewable energy over the long term which was found to be

around $0,025/kWh [22] which is based on long term purchase

power agreements (PPA) for wind farms in place today through

the year 2040. Secondly, the ERCOT data is based on

wholesale energy costs and not delivered costs to the customer.

Data from the US EIA was found to show the average retail rate

for residential applications for the state of Texas to be

$0.1188/kWh [23] so a multiplier was developed to bring the

average wholesale energy cost for the year equal to the EIA

data. This information is summarized in Table 2.

Table 2. Variables Needed for Rate Setup

Variable Type Amount

Renewable Energy Base Cost [22] $0.025 /kWh

Renewable Gen. Percent of capacity [21] 33%

Multiplier; wholesale to retail [18], [23] 3.12

With this information in place, the fifteen minute rate

structure for the year can be cal calculated to include both

traditional and renewable energy costs as well as incorporating

a proxy for the amount of demand on the grid through the

ERCOT wholesale energy cost data. This information is shown

in Figure 6 where the red line represents the monthly load

shape of the renewable energy output as a percentage in March

(left axis) and the blue line is the corresponding monthly

average energy costs for each period of the day (right axis).

This information shows a strong correlation between renewable

energy generation and energy costs. However, it is not absolute

as there are locations on the graph where large amounts of

energy is being generated, but due to the ERCOT data, it is

known that the demand for energy is also high, therefore the

cost does not follow.

Figure 6. Profile of Energy Cost and Renewable

Generation Load Shape for March

The next source of data necessary to model the integration

system is a baseline load shape pattern for a home which will

be used to determine feasible times throughout the day where

appliances can operate without imposing an inconvenience to

the homeowner and to also account for human behavior in

energy consumption. The major source for this data comes

from a large load research project undertaken in the Pacific

Northwest region of the US, starting in April 2012 and running

through July 2014 by Ecotope, Inc. and funded by the

Northwest Energy Efficiency Alliance (NEEA) and Bonneville

Power Administration (BPA) [24]. This study captured fifteen

minute energy usage data for all types of electrical appliances

within 101 homes and was made publically available through

the original project.

For the purpose of this project a subset of homes were

randomly selected to be utilized. It should be noted that while

the information is valuable, the climate for the Pacific

Northwest is very different than the one in the Southeast US,

therefore only non-weather dependent end uses were

considered. For example, clothes washer and dryer, oven and

dishwashers were included but HVAC and water heating were

excluded and will be included in the model with the use of

Renewable

Energy Costs


thermostat changes rather than a fixed load shape. Additionally

other appliances are seen as non-interruptible such as a TV, PC

or lighting and are not controlled by the home controller either.

The interval end use load data and the interval cost data

discussed in this section are combined through the use of an

algorithm which takes the daily profiles and searches for the

combination of the lowest cost and the highest probability that

the customer would operate each appliance during that time

span.

FUTURE RESEARCH The research work presented in this paper is a foundation

that will be expanded upon under the same project. The next

step in modeling will include developing and automating the

process and algorithm by which the optimal time to schedule an

appliance to operate is found for the corresponding day of the

year. A separate rate structure will be developed based off the

one presented previously that will incorporate errors in the

forecasting methodology which will then be presented to

energy storage devices within the home to compensate. The

response (charge and discharge) will need to be developed to

optimize the storage potential and will be done by changing

temperature setpoints, charge rates, etc.

Finally all the controls and pricing schemes will be

modeled together with approximately ten homes at once. This

system model will be developed utilizing the Building Controls

Virtual Test Bed (BCVTB), a software environment which

allows for co-simulation between different software packages

[25]. The BCVTB tool will be utilized to link, in real-time,

EnergyPlus and Matlab and allow them to exchange relevant

information to incorporate the necessary short-term

optimizations while also providing an actual building

simulation to model the day-ahead schedules and determine

how changes within the home impact energy consumption and

comfort as a whole and how this can be used to optimize the

grid while reducing the energy costs for the consumer.

This simulation model will also allow for an iterative

approach where the energy consumption and load shape results

from a simulation will be fed into the beginning of the next

iteration to simulate how energy consumption changes in

response to updated energy costs.

SUMMARY The research project presented focuses on the day-ahead

time frame optimization of the grid based on traditional

forecasting techniques of energy demand along with methods of

how energy will be generated. In addition, a shorter time

horizon response to utilize inherent thermal energy storage

available on the grid as well as electrical energy storage to

compensate for longer term errors that arise will be utilized.

This dual time scale and technology focus is meant to address

gaps in current research and to be utilized by a utility provider

to allocate generation resources more effectively and increase

the overall efficiency of the grid while also being able to

compensate for errors in the longer term renewable energy

output and demand side forecasts. This approach is not meant

to address all issues related to integrating renewable energy

such as frequency response which must be performed in the

sub-second time frame but is meant to serve as an optimization

technique and improve the link between energy usage and

renewable energy generation, bridging a gap in today’s grid to

allow for higher levels of renewable energy to supply the

electricity needs of the country while also minimizing or even

eliminating the amount of renewable energy generation that

must be curtailed due to lack of demand side optimization.

REFERENCES

[1] C. Ashley, L. Holmes and G. Wikler, "Using More

Energy Can Be a Good Thing: C&I Loads as a Balancing

Resource for Intermittent Renewable Energy," in ACEEE

Summer Study on Energy Efficiency in Industry, Niagara Falls,

2013.

[2] S. Li, D. Zhang, A. Roget and Z. O'Neill, "Integrating

Home Energy Simulation and Dynamic Electricity Price for

Demand Response Study," IEEE TRANSACTIONS ON SMART

GRID, vol. 5, no. 2, pp. 779-788, 2014.

[3] P. Samadi, H. Mohsenian-Rad, V. W. Wong and R.

Schober, "Utilizing Renewable Energy Resources by Adopting

DSM Techniques and Storage Facilities," IEEE ICC , pp. 4221-

4226, 2014.

[4] X. Chen, W. Tongquan and S. Hu, "Uncertainty-Aware

Household Appliance Scheduling Considering Dynamic

Electricity Pricing in Smart Home," IEEE Transactions on

Smart Grid, pp. 932-941, 2013.

[5] A.-H. Mohsenian-Rad, V. W. Wong, J. Jatskevich, R.

Schober and A. Leon-Garcia, "Autonomous Demand-Side

Management Based on Game-Theoretic Energy Consumption

Scheduling for the Future Smart Grid," IEEE TRANSACTIONS

ON SMART GRID, pp. 320-331, 2010.

[6] J. H. Yoon, R. Baldick and A. Novoselac, "Dynamic

Demand Response Controller Based on Real-Time Retail Price

for Residential Buildings," IEEE TRANSACTIONS ON SMART

GRID, pp. 121-129, 2014.

[7] V. Bakker, M. Bosman, A. Molderink, J. Hurink and

G. Smit, "Demand side load management using a three step

optimization methodology," in SmartGridComm, Gaithersburg,

2010.

[8] D. Li and S. K. Jayaweera, "Distributed Smart-Home

Decision-Making in a Hierarchical Interactive Smart Grid

Architecture," IEEE TRANSACTIONS ON PARALLEL AND

DISTRIBUTED SYSTEMS, vol. 26, no. 1, pp. 75- 84, 2015.

[9] M. H. Yaghmaee, R. Minoochehr and A. Saeedi,

"REAL TIME DEMAND RESPONSE USING RENEWABLE

ENERGY RESOURCES AND ENERGY STORAGE IN

SMART CONSUMERS," in CIRED, Stockholm, 2013.

[10] P. Samadi, H. Mohsenian-Rad, V. W. Wong and R.

Schober, "Tackling the Load Uncertainty Challenges for Energy

Consumption Scheduling in Smart Grid," IEEE Transactions on

Smart Grid, pp. 1007-1016, 2013.

[11] A. Molderink, V. Bakker, M. G. Bosman, J. L. Hurink

and G. J. Smit, "Improving stability and utilization of the


electricity infrastructure of a neighborhood," in CITRES,

Waltham, 2010.

[12] A.-H. Mohsenian-Rad and A. Leon-Garcia, "Optimal

Residential Load Control With Price Prediction in Real-Time

Electricity Pricing Markets," IEEE Transactions on Smart Grid,

vol. 1, no. 2, pp. 120-133, 2010.

[13] S. Yue, J. Chen, Y. Gu, C. Wu and Y. Shi, "Dual-

pricing Policy for Controller-side Strategies in Demand Side

Management," in IEEE SmartGridComm, Brussels, 2011.

[14] National Renewable Energy Lab, "BEopt Home,"

2015. [Online]. Available: http://beopt.nrel.gov/. [Accessed 22

August 2015].

[15] US Department of Energy, "EnergyPlus Energy

Simulation Software," 2015. [Online]. Available:

http://apps1.eere.energy.gov/buildings/energyplus/energyplus_a

bout.cfm. [Accessed 22 August 2015].

[16] Iowa State University of Science and Technology,

"ASOS-AWOS-METAR Data Download," 2015. [Online].

Available:

http://mesonet.agron.iastate.edu/request/download.phtml?netwo

rk=AL_ASOS. [Accessed 22 August 2015].

[17] SolarAnywhere, "Data," Clean Power Research, 2015.

[Online]. Available:

https://www.solaranywhere.com/Public/SelectData.aspx.

[Accessed 22 August 2015].

[18] Electric Reliability Council of Texas, "Market Prices,"

Electric Reliability Council of Texas, 2005. [Online]. Available:

http://www.ercot.com/mktinfo/prices. [Accessed 11 August

2015].

[19] NREL, "Solar Power Data for Integration Studies," 26

May 2015. [Online]. Available:

http://www.nrel.gov/electricity/transmission/solar_integration_

methodology.html. [Accessed 15 August 2015].

[20] NREL, "Wind Prospector," 2015. [Online]. Available:

https://maps.nrel.gov/wind-prospector/. [Accessed 13 August

2015].

[21] California Public Utilities Commission, "California

Renewables Portfolio Standard (RPS)," 2007. [Online].

Available:

http://www.cpuc.ca.gov/PUC/energy/Renewables/index.htm.

[Accessed 28 January 2014].

[22] US Department of Energy EERE, "2014 Wind

Technologies Market Report," EERE, DC, 2015.

[23] US EIA, "Electric Power Monthly," May 2015.

[Online]. Available:

http://www.eia.gov/electricity/monthly/epm_table_grapher.cfm

?t=epmt_5_6_a. [Accessed 15 August 2015].

[24] Ecotope Inc., "Residential Building Stock Assessment:

Metering Study," 28 April 2014. [Online]. Available:

http://neea.org/docs/default-source/reports/residential-building-

stock-assessment--metering-study.pdf?sfvrsn=6. [Accessed 8

August 2015].

[25] Lawrence Berkeley National Labs, "Building Controls

Virtual Test Bed," Lawrence Berkeley National Labs, 31

January 2015. [Online]. Available:

https://simulationresearch.lbl.gov/bcvtb. [Accessed 25 August

2015].

[26] M. H. Yaghmaee, R. Minoochehr and A. Saeedi,

"REAL TIME DEMAND RESPONSE USING RENEWABLE

ENERGY RESOURCES AND ENERGY STORAGE IN

SMART CONSUMERS," in 22nd International Conference on

Electricity Distribution, Stockholm, 2013.




SOLAR SIMULATOR MIXED TEMPERATURE TEST OF WIRE MESH PARTICLE HEATING RECEIVER TO MEASURE RECEIVER EFFICIENCY

Matthew Golob, Clayton Nguyen, Sheldon Jeter, Said Abdel-Khalik

Georgia Institute of Technology Atlanta, Georgia, USA

ABSTRACT

Solar receiver technology is exploring new possibilities,

particularly in pushing higher temperature ceilings to improve

power cycle efficiency. This move has shifted solar receivers

from troughs using heat transfer oils to central receivers

utilizing molten salts that can capitalize on the higher

concentration flux and resulting higher output temperature.

While molten salts have allowed for high receiver temperatures,

they are limited by cost and composition. The higher

temperatures and potentially resulting higher efficiency can be

reached by utilizing particulates instead. Particulates such as

commercially available aluminia based products offer two key

benefits. (1) Unlike typical molten salts, which are restricted

by phase changes to operating ranges between 240°C to 565°C

[1], these ceramic particles are stable to over 1000°C [2] and

ID50-K melt at 2200°C [3]. (2) While not a fluid, they can still

be flowed through a structure allowing for much better heat

exchange regimes.

To explore this alternative potential, an experimental trial

at Georgia Tech has been further developed [4] to assess the

efficiency and effectiveness of a heating receiver employing

particles. The basic approach of the receiver is to drop particles

vertically through the irradiated space in order to heat them.

Instead of simply dropping the particles through the heated

zone, a chevron wire mesh is employed in the receiver to slow

the free fall of particles, increasing residence time and

temperature rise per fall length. The overall experiment uses a

calibrated source of concentrated radiation from a solar

simulator as the energy input and measures the energy

collection from the temperature rise in a mass flow rate of

particulate dropped through the test receiver. With a measured

energy rise of the particulate and known energy input, the

particle heating receiver efficiency can be calculated.

The results here are from data collected using a small scale

receiver employing an improved mixer stage to achieve a better

exit flow measurement. The apparatus utilizes the Georgia

Tech Solar Simulator, which uses a bank of high intensity

xenon lamps to simulate a concentrated solar source. These

results show the thermal efficiency of a small scale particle

heating receiver from 35-160ºC to be around 85-92%.

INTRODUCTION

The experiment covered in this paper is an ongoing

iteration of test receiver designs utilizing Georgia Tech’s Solar

Simulator (GTSS). What is learned here is used to support key

components of a larger-scale receiver design involving the

Department of Energy’s SunShot program. The GTSS consists

of a bank of 7 xenon lamps, all focused down to a point

approximately 80 mm in diameter, Figure 1. This device serves

as a convenient artificial concentrated solar irradiation source

and can output concentration ratios well in excess of 1000 suns.

Figure 1. GTSS test of Focal Plane

Figure 2. Former recirculating OLDS elevator

configuration

Improvements on the previous particle heating receiver

(PHR) apparatus resulted in a switch from a recirculating


OLDS elevator configuration, Figure 2, to a large single pass

hopper particle system for better temperature stability and mass

flow sensing. The previous setup also tested likely receiver

designs, covering two receiver configurations: a simple free-

falling curtain and a chevron wire-mesh design to inhibit free

flow. These will be repeated and further studied. The GTSS

has also undergone considerable improvements to correct some

electrical issues as well as refinements in the operation,

alignment, and focus of the lamps. As a result a more nearly

uniform (but not perfect) hot spot can now be generated with

about 80% of the incident radiation falling within an

approximately 80 mm diameter circle. The updated apparatus

can simulate the high fluxes expected in practical operations

(~250 to as much as 2000 kW/m2).

The current apparatus has been developed to provide a

single-pass high-temperature transient test employing the

GTSS. This apparatus consists of a 25.4 L supply hopper that

employs a knife valve to remotely release particles through the

PHR test unit. Another identical 25.4L hopper is set up on a

scale to capture the particles that pass through the PHR test

unit. The hoppers are interchangeable and are designed for

repeated use, allowing for the pre-heated particles of a previous

run to be used as the inlet particles for a subsequent run through

switching the locations of hoppers after a test. During testing a

fixed flow rate into the irradiated test receiver region located at

the focus, Figure 3, thereby simulates the operation of a small

representative subset of a larger PHR. The flow rate is regulated

utilizing a perforated plate, which simultaneously disperses and

controls the flow to prevent regional overloading in the receiver

as well as ensuring a constant flow rate in the saturated state.

Figure 3. Receiver Test Region

An important addition to the PHR test unit assembly been a

mixer section. In the prior experimental iteration [4], local

differences in the flux exposure caused both a lateral and depth

wise gradient in the resulting mass flow of heated particles. To

more clearly determine the overall PHR efficiency, it was

deemed necessary to establish a mixed temperature of the

exiting PHR flow. The solution came in the form of a series of

linearly converging funnels, with each stage oriented 90º off

from the one above (see Figure 4). The results show that four

of these stages could produce a well-mixed exit flow

temperature particularly if the particle inlet temperature to the

PHR was uniform. Further measurement improvements include

a better thermocouple placement and catch design for getting a

clean particle saturated flow temperature. The results of this

setup to measure PHR efficiency are what will be reported in

this paper.

Figure 4. Model of two of the mixer stages

LITERATURE REVIEW One purpose of the SunShot project is to explore and

improve the particle heating receiver component of the solar

concentrator power cycle. This is a relatively new field of

research, so there are a limited range of studies with which to

compare.

Previous research at SNL [5] has examined alternative

receiver designs for a high efficiency particulate solar receiver.

These designs feature a receiver cavity box with a free falling

particulate curtain. Their work focused on varying the cavity

depth, the ceiling slope angle, the specular properties of the

walls and the back geometry in order to improve the efficiency

of the solar receiver. In comparison to their base receiver design

with a vertical aperture, the new design increases the theoretical

thermal efficiency from 72.3% to 86.8%. SNL’s work on the


free falling curtain is a core PHR candidate for the SunShot

program. The main drawback is the significant particle

acceleration of the free fall setup, a downside of which the

chevron mesh design seeks to eliminate.

Tan et al. [6] also looked at falling particle receivers. Their

research mainly dealt with computational simulation of wind

effects on the receiver. An aerowindow was a suggested

addition to the aperture of solid particle solar receivers. The

aerowindow acts as an air curtain which would help prevent the

loss of heated air to the ambient conditions. In addition, it has

the added advantage of ensuring particle retention. While

noteworthy ideas, there is little here in terms of physical tests or

measurements against which to compare receiver efficiencies.

Xiao et al. [7] is perhaps one of the closest experiments to

ours, mainly through the use of lamps and physical receiver.

Zhejiang University used a similar Xenon-arc lamp bank as a

solar simulator to test a spiral solid particle solar receiver. This

receiver cavity resembled a sunken helical spiral and employed

a top facing aperture with a glass window cover. The receiver

was experimentally measured to achieve a temperature rise

exceeding 350°C for a single pass with a 19.3 kW/m2

focal flux

off the lamps. The receiver had an optical efficiency of 84%

and a thermal efficiency of 60%. The general setup is somewhat

similar to our test, although there are significant differences in

receiver geometry, orientation, and particle flow regimes.

Rӧger et al. [8] looked at different Solid Particle Receiver

(SPR) designs utilizing varying particle recirculation schemes

in order to maximize the particle heating. The general particle

recirculation schemes focused on increasing particle residence

time to in turn increase the thermal efficiency of the receiver.

While the studies do not rigorously account for convective

losses, they did highlight the need to increase particle residence

time in the receiver.

EXPERIMENTAL SETUP

The chevron mesh design to be tested is also similar to the

inlet region of a falling curtain in terms of average particle fall

speed. As an objective, the ultimate form of the system will be

able to investigate high-temperature collection efficiency and

provide empirical data to support detailed computer modeling.

The basic design for the system can be seen in Figure 5. The

cone of light shown in the image illustrates a representative

inlet cone from the seven solar simulator lamps. To achieve a

suitable efficiency determination, three key measurements had

to be taken. First the quantity of thermal input into the PHR

cavity from the lamps had to be established. This was done

using a water cooled calorimeter. Next the mass flux rate of

particles passing through the PHR needed to be recorded. This

was accomplished by controlling the inlet area and capturing

the mass flow via scale measurements. The remaining critical

measurement is of the temperature change across the PHR test

section, where the addition of a mixer stage was important in

establishing a clean particle exit temperature. The calorimeter,

PHR mass flux, and PHR temperature difference were the key

parameter measurements to be set up for this experiment.

Figure 5. Single pass test apparatus

Calorimeter

To determine the incident solar irradiation the GT SunShot

Research Team has been assisting the GTSS operators in

developing a water cooled cavity calorimeter, similar to the one

produced by Groer and Neumann [9], to accurately measure the

amount of concentrated radiation being delivered to the

receiver. For the mixed temperature PHR tests conducted with

the single pass receiver, one and three lamps were used.

Figure 6: Calorimeter on Test Stand with Iris Plate

The calorimeter consists of a copper surface coated in high

absorptivity black. When an iris plate is mounted to the front of

the calorimeter, the core serves as a black body, Figure 6. The

core is further swathed in insulation to minimize heat loss. The

assembly is encased in a large steel pipe to provide support as

well as mountings for the iris plate. Line water is run through

copper tubing in the assembly with temperature taps taking the


incoming and outgoing water stream measurements. A positive

displacement flow meter is also located on the water line to

gauge the water flow rate. The resulting heat inputs of the

lamps are determined by the following equation.

in,out,wp,wInput ww TTcmQ

(1)

where Input

Q is the calculated heat rate from the lamp, wm

is

the recorded flow rate of the line water, wpC , the specific heat

of water, outwT , the measured outlet water temperature, and

inwT , measured the inlet water temperature. In order to apply a

lamp’s heat rate to a different experimental setup, an image of a

Lambertian target at the same focal plane as the calorimeter

was taken. The intensity of the light recorded in the image was

then calibrated to the measured heat rate for that plane at the

diameter of the calorimeter iris. With this and a new image of

the PHR as well as the new iris diameter on the receiver, an

equivalent heat input rate could be correlated to the test wire

mesh PHR. At that time the average of the best apparent image

data and available calorimetry measurements gives a heat rate

of ~4.7 kW into the PHR from three lamps and ~1.7kW for one

lamp operation.

PHR Mass Flux Rate

In order the accurately measure the PHR efficiency it was

critical to determine and control the mass flux passing into the

PHR cavity. To accomplish this, a scale base was characterized

and calibrated over the range of mass available for the PHR

tests. Once calibrated, the scale was used in conjunction with a

series of porous plates to measure the rate of change of the

mass accumulating within the hopper that catches all of the

PHR discharge. To meet the target mass flux goals the porous

plates were modified to particular open area ratios yielding

differing mass flow rates over the same constrained flux area.

The resulting mass flow rate was determined from analysis of

the transient mass accumulation recorded by the scale base.

This, in conjunction with the fixed flux area, allowed the PHR

mass flux to be established.

Base Scale Calibration

Mass calibration of the scale base used to measure the

mass flux rate was done incrementally using an existing hopper

filled with ID50-K particulate and an empty hopper set on a

scale base mounted in the planned PHR test configuration. A

beaker full of ID50-K was then taken out of the full hopper

placed on an OHAUS GT8000 (±0.1g) scale, where it was

weighed and recorded with the beaker’s mass tared out, then

poured into the empty hopper. This process was repeated until

the empty hopper on the scale base was filled to ~75 kg. A few

extra weights were added to increase the calibration range in

case of future modifications to the hopper. This yielded a

calibration for the scale base accurate to within ±2.5 g and

±0.21g/s transiently over the needed test range.

Figure 7. Controlled Perforated Discharge Area

A funnel guide was used to constrain the discharge area of

the perforated plate. The holes within the perforated plate were

controlled to an open area (OA) ratio of 41.5% with pattern of

0.156” diameter openings to meet task mass flux target. The

source perforated plate had 0.156” diameter holes with 0.219”

spacing for an open area ratio of 46%, Figure 7. Seven holes

were blocked to get the open area ratio to 41.5% which yielded

a suitable mass flux rate.

To keep the receiver tests consistent between the large

scale and small scale simulator tests, the particle mass flux of

the GTSS is targeted to be between 65 and 75 kg/m2-s. The

GTSS setup uses the manually calibrated scale base load cell in

order to measure the mass in the outlet hopper. Post processing

is then used in order to transiently measure the mass flow and

flux rates. The rate of change in the load cell signal is

calculated using a five point stencil as shown below,

avg12

218

18

2

t

iiii

dt

id

(2)

where tavg represents the average time differential between each

measurement, ε is the signal voltage and i is the index for each

measurement. A calibration constant Km is then used to convert

the data into a mass flow rate.

Cal

Kdt

dm

(3)

The length, L and width, W of the inlet is used to calculate the

mass flux rate, ṁ” through the receiver.

LWmK

dt

dm

1

(4)

By taking a graph of the accumulated mass against the time one

can see that the average mass flow rate is 0.13 kg/s with an R2

value well above 0.90, indicating a steady mass flow rate,

Figure 8. This lead to an average mass flux rate of 67.68 ± 0.39

kg/m2-s.


Figure 8. Mass Accumulation over Time for 41.5% OA

Utilizing the five point stencil the mass flux rate can be

transiently calculated for each measurement. For the mass flux,

the open area ratio of 41.5% with an average mass flux of 67.68

kg/m2-s fell within the SunShot project target flow parameter

of 65 to 75 kg/m2-s.

PHR Temperature Capture

As had been noted in earlier iterations of this solar

simulator PHR experiment [4] there were issues with capturing

a clean PHR exit temperature. The previous method used a

mesh with a grid of thermocouples to get a characterization of

the exit flow. The non-uniformity of the grid thermocouples

brings issues in how best to characterize weight each due the

local mass flow rates. An answer was to instead mix the exiting

stream, getting the complete mass flow rate of particles leaving

the PHR at a uniform mixed temperature.

Figure 9. Particle mixing in a linearly converging funnel

stage

Unlike true fluids, mixing particle flows is fairly difficult.

Particle flows tend to behave in an extremely laminar fashion,

with little turbulent mixing between layers. The employed

mixing solution came in the form of a linearly converging

funnel stage. This constriction allows two captured streams of

flowing particles coming off the bottom of the funnel to freely

collide and turbulently mix, Figure 9. Each stage allows

mixing along one axis. By rotating the subsequent stage by 90º

a more thorough mixing was achieved. Testing showed that

after four stages, particles with a profile temperature similar to

that passing through an irradiated PHR would have a uniform

temperature in the exit stream.

Figure 10. Looking from underneath up at mixing stages

and wire mesh thermocouple catches

To measure the flow exiting the mixing stages, a set of

wire mesh thermocouple flow catches were designed, Figure

10. These catches ensured that the thermocouple beads remain

submerged in particle flow avoiding disruptive air influences

while still allowing for flow bypass to avoid choking and

stoppage issues. Overall this setup in conjunction with

thermocouples located on the controlled perforated discharge

area, Figure 7, permitted suitable temperature readings across

the PHR test unit.

PHR Test Unit

To test the core PHR design concept, a small scale receiver

was built for experimenting in the Solar Simulator Lab. The

receiver’s back wall is 0.102 m by 0.203 m and acts as a

representative portion of a larger receiver that will be used at

SNL. The receiver space is filled with 10 mesh wire chevrons

that slow the falling particles in the irradiated zone. The

targeted focal plane is an inch off the back wall towards the

receiver aperture. According to the GTSS team’s simulator

modeling, that spot should receive approximately 80% of the

irradiative power provided by the lamps. The iris plate that

covers the front of the receiver was built to allow the 80%

portion of the light in while shielding rest of receiver from the

remaining incident irradiation. This also protects the

thermocouples near the receiver from any direct irradiative

exposure.


Figure 11. Solar simulator test PHR with water cooled

shield

The single-pass solar simulator is being tested using ID50-

K particulates. The ID50-K is a primarily alumina comprised

particulate [3] that has high absorptivity and is the candidate

medium for the SunShot PHR. In order to run this test, a water-

cooled Lambertian shield was placed in front of the receiver to

protect it until the lamps reached steady state operation, Figure

11. A few seconds prior to the shield’s retraction, the valve

controlling the particulate flow would be opened to allow for

ID50-K to start passing in a steady flow state through the

receiver. This would begin a test run. The run ends when the

top hopper is nearly exhausted.

MEASUREMENTS Three K-type Thermocouples with a ≥±0.2°C accuracy are

placed into the perforated discharge area to measure the

particulate temperature as it enters the PHR. Four more K-type

thermocouples (also ≥±0.2°C) with wire mesh catches are set at

the bottom of the mixer stages downstream of the PHR to

measure the exiting particle stream temperature. As testing

proceeded it was discovered that a growing temperature

gradient experienced with the hopper inlet particulates makes it

difficult to measure a stable temperature difference across the

PHR at increasingly elevated temperatures.

The efficiency of the receiver was calculated by comparing

the assessed input heat rate from the lamp, Input

Q , to the

measured heat rate gain seen in the particulates, Part

Q

in,out,pp,pPart pp TTcmQ

(5)

The heat rate gain of the particulate is calculated by, , which

is measured mass flow rate of the particulate into the base

hopper, p,pc , the specific heat of ID-50K, out,pT , the average

mixed stream temperature at the receiver discharge or mixed

hopper temperate, and in,pT , the average incoming particulate

stream temperature from the top hopper. The resulting

efficiency is calculated by equation 6.

Input

Partr

Q

Q (6)

where the calculated receiver efficiency is the heat rate gain of

the particulate divided by the heat input rate from the lamp.

Solar Simulator PHR Efficiency Testing The Solar Simulator Lab was used to test the small scale

PHR for a for a series of 5 runs with the average temperature

ranging between 35°C to 160°C. The current test apparatus is

an updated iteration replacing the temperature limited Olds

Elevator recirculation method with a single pass insulated

hopper design. These tests were run consecutively using the

batch processing setup to gather receiver efficiency values for

progressively higher temperatures.

The first run is used to calculate the optical efficiency; as

such this test was conducted using only one lamp,

approximately 430 suns. The following five runs were

conducted using 3 lamps, approximately 1050 suns. Subject to

the chosen specific heat, the efficiency for the runs with stable

temperature differences (the initial two in the table had PHR

efficiency’s between 85-92%.

Table 1. Receiver efficiency values for different specific heat sources

The range of calculated efficiencies was dependent on the

choice of various experimental and theoretical methods for

finding the specific heat, Figure 12. SNL specific heat was from

a test conducted using a NETZCH Simultaneous

Thermogravimetry - Differential Scanning Calorimeter (DSC).

The semi-empirical calculation employed IUPACK and NIST

data along with application of the Kopp-Neumann material

compositional law for the specific heat. Clemson University

(CU) had also conducted 4 runs on ID50-K using a NETZCH

Differential Scanning Calorimeter (DSC). The Clemson

specific heat is based off an empirical polynomial regression

model of the four different runs. The DSC used at CU was also

used to measure the specific heat of alumina powder for use as

a calibration standard on the device. This uncertainty was then

used to find the uncertainty of the specific heat measurements

of the data from CU for the ID50-K.


Figure 12. Specific heat of ID50-K from using NETZSCH

DSC 404C, NETZSCH STA 409 C/CD and Kopp-Neumann

The first two runs had the most reliable measurements due

to the uniform ambient temperature of the ID50-K in the inlet

hopper. Each efficiency is calculated assuming no uncertainty

in the specific heat model. Consecutive runs have shown that

while an accurate mixed outlet temperature can be measured,

the inlet hopper naturally develops a thermal gradient due to the

ambient conditions of the outlet hopper. As the tests progressed

this lead to progressively larger thermal gradients resulting in

unreliable inlet temperature measurements, which contributed

to the relatively high receiver efficiencies in later runs.

Table 2. Temperature data for the different runs, showing progressively larger non-uniformity at the inlet

Over the consecutive runs the data shows that the change

in average particle temperature begins to decrease between

runs. This is because the valve that is used creates a thermal

leak as well as the thermal mass of the hopper itself. The heat

leak in the experiment, in addition to the non-uniform

temperatures caused by these leaks makes it unrealistic to use

repeated batch runs to reach higher temperatures.

CONCLUSION

The preliminary results of the test show that the discrete

structure receiver using a size 10 mesh can potentially achieve

receiver efficiency greater than 85-92% depending on what

specific heat is used. However, to test the PHR at higher

temperatures the inlet reservoir will need to be preheated to

500°C using a combination of band and air/mixer heater. These

heaters will be kept at a constant temperature using a solid state

relay with PID control over a prolonged period of time. To

ensure a uniform temperature profile within the hopper

reservoir, an air ejector is being modified for use as a small

scale non-mechanical particle mixer. The air that is used within

the ejector will also be preheated to prevent overly cooling the

particles. With these changes, upcoming tests will further focus

on repeating runs at elevated temperatures and comparing runs

with altered mesh layouts as well as a free falling configuration.

Overall these measurements will determine the PHR efficiency

over a wider range of conditions and guide the design path of

the SunShot’s large scale PHR.

ACKNOWLEDGMENT Financial support of the US Department of Energy through the

SunShot research program is recognized and appreciated.

REFERENCES [1] Ortega, J., Burgaleta, J., Téllez, F., 2008, “Central receiver

system solar power plant using molten salt as heat transfer

fluid”, Transactions of the ASME, 2008, Journal of Solar

Energy Engineering, 130.

[2] Hruby, J. M., 1986, "A Technical Feasibility Study of a

Solid Particle Solar Central Receiver for High Temperature

Applications,” SAND87-8005, 1986, Sandia National

Laboratories Albuquerque, NM

[3] “CarboAccucast ID50-K”, Carbo Ceramics, 2014, Houston,

TX, www.carboceramics.com

[4] Golob, M., Nguyen,C., Gill, R., Jeter, S., Abdel-Khalik, S.,

2014, “Solar Simulator Test of Wire Mesh Particle Heating

Receiver to measure receiver efficiency”, Early Career

Technical Conference, 2014, Birmingham, AL

[5] Christian, J., Ho, C.K., 2013, “Alternative Designs of a

High Efficiency, North-facing, Solid Particle Receiver”,

SolarPACES 2013, Energy Procedia 49 (2014) pg. 314-324,

doi: 10.1016/j.egypro.2014.03.034

[6] Tan, T., Chen, Y., Chen, Z., Siegal, N., Kolb, G., 2009,

“Wind effect on the performance of solid particle solar

receivers with and without the protection of an aerowindow”,

Solar Energy, Oct. 2009, Vol. 83, pg. 1815-1827

[7] Xiao, G., Guo, K., Luo, Z., Ni, M., Zhang, Y., Wang, C.,

2014, “Simulation and experimental study on a spiral solid

particle solar receiver”, Applied Energy, Jan. 2014, Vol. 113, pg.

178-188

[8] Roger, M., Amsbeck, L., Gobereit, B., Reiner, B., 2011,

“Face-Down Solid Particle Receiver Using Recirculation”,

Journal of Solar Energy, July 2011, Vol. 133,

doi:10.1115/1.4004269

[9] Groer, U., and Neumann, A., 1999, "Development and test

of a high flux calorimeter at DLR Cologne," Journal De

Physique. IV: JP, 9(3), pp. Pr3-643 - Pr643-648.

[10] Defoe C. Ginnings and Robert Corruccini, 1947,

“Enthalpy, Specific Heat, and Entropy of Aluminum Oxide

from 0° to 900° C”

http://www.carboceramics.com/


Documents

SECTION 4 - UAB · 2018-07-24 · Ketan Solanki, Mohamed Nizamuddin Shaik, Vijaya Krishna Teja Bangi Department of Mechanical Engineering Lamar University Beaumont, TX, USA Dr. Ramesh