Study - Axess Systems · Ecological comparison of PC and thin client desk top equipment compiled for: IGEL Technology GmbH Mr. Heiko Gloge Schlachte 39/40 28195 Bremen / Germany Oberhausen,

Study

Ecological comparison of PC and thin client desk topequipment

compiled for:IGEL Technology GmbHMr. Heiko GlogeSchlachte 39/4028195 Bremen / Germany

Oberhausen, 18th December 2006

Study

Ecological comparison of PC and thin client desk top equipment

compiled by:Fraunhofer-Institut für Umwelt-, Sicherheits-und Energietechnik UMSICHTInstitute DirectorProf. Dr.-Ing. Eckhard WeidnerOsterfelder Straße 346047 Oberhausen / GermanyProject team:Dipl.-Inform. (FH) Christian Knermann | Project manager

IT ManagementTel.: 02 08/85 98-11 18E-Mail: [email protected]

Dr.-Ing. Hartmut PflaumResource management division managerTel.: 02 08/85 98-11 71E-Mail: [email protected]

Dipl.-Wirt.-Ing. (FH) Markus Hiebel (MSc) | deputy project managerProject managerResource management divisionTel.: 02 08/85 98-11 81E-Mail: [email protected]

Manuela Rettweiler, M.A.Marketing, Communication, Corporate planningTel.: 02 08/85 98-14 12E-Mail: [email protected]

Dr.-Ing. Ulrich SeifertTechnical production information systemsTel.: 02 08/85 98 – 11 27E-Mail: [email protected]

Copyright

The copyright for the designs, drafts, analyses, studies and other documents created forthis study by Fraunhofer UMSICHT belongs to Fraunhofer UMSICHT. Any transfer ofcopyright must be in writing.The client is entitled to use this study for purposes provided for in the order. Reproducingthis study is only permissible with the express approval of Fraunhofer UMSICHT. Alterations,translations or digital post processing are not permitted. It is not permitted to pass thestudy on to a third party, particularly competitors of Fraunhofer UMSICHT – with the

exception of public subsidy providers or investors - without the written permission ofFraunhofer UMSICHT.© Copyright Fraunhofer UMSICHT 2006

I

Contents

1 Abstract (executive summary) 1

2 Nature of the task 3

3 Method 33.1 Definition of scenarios 33.1.1 PC systems 43.1.2 Thin Client systems 53.2 Legal position 103.3 System limits and material streams 113.4 Selecting a suitable instrument for the ecological

comparison of work station equipment 133.5 Selection of characteristics for comparing work station

equipment 153.5.1 Production phase 153.5.2 Use phase 163.5.3 Disposal phase 163.6 Data collection procedure 173.6.1 Weight 173.6.2 Power consumption 183.7 Comparison of workstation equipment, validation and

sensitivity analyses 19

4 Summary and results from the use scenarios 204.1 Introducing the scenarios 204.2 Results 214.2.1 Production phase 214.2.2 Use phase 264.2.3 Disposal phase 324.3 Summary of the results (comparison of workstation

equipment) 334.4 Validation and sensitivity analyses 35

5 Interpretation of the results 37

6 Sources 40

1Ecological comparison of PCs and thinclients

18th December 2006

1 Abstract (executive summary)

In this project a comparison was made between the environmental effects of a PCand a thin client-supported provision of IT services. In this comparison themanufacturing phase, the use phase and the disposal phase were taken intoconsideration using a modified assessment instrument (ecological performanceindicators) (cf. sections 2 and 3.4).

In order be able to compare PC and thin client desk top equipment , first of all"equivalent" PC and thin client systems were identified which took intoconsideration the different "user types" (intensity of use). Here a model consistingof the three user groups, Light User, Medium User and Heavy User was used, ashad already been used in the previous economic study [UMSICHT, 2006]. Theseuser groups were allocated terminals depending on their requirements. Here onlythe hardware which was directly connected with one or other operating modelwas examined (cf. section 3.1.1).

Only the direct effects of the individual product life phases were examined. Theprevious chains, i.e. the extraction of metals and the production of fuels are notsummed up individually. After production the desk top equipment is distributedand operated in the companies. During the use phase the equipment needsenergy (power) and creates emissions e.g. in the form of noise. When they havefinished being used the units are collected and disposed of or recycled. Thedisposal phase and thus the life cycle studied finishes here. Secondary data(literature), modelling calculations and our own measurements in the laboratorywere used to collect the data (cf. sections 3.3 and 3.6).

The comparison can only extend to application profiles where the performancerange of the thin client system is comparable with that of a PC system. In mostcases this is currently fulfilled if mainly standard office applications and, to a lesserextent, special applications (e.g. graphics editing) are required. Also the thin clientcannot be customised (e.g. by installing additional equipment like drives, TV cardsetc.). If the level of customisation or extension is too great the advantages of thehitherto standard orientated thin client concept will not be so great.

Nature of the task

Procedures andmethods


18th December 2006

Table 1: Qualitative summary of results (for details see the specified sections)

Factor/criterion Thin Client compared

with the PC is...

Remarks see

section

Transport/Logistics ++ smaller and lighterhigh levelof transport density

lower emissions per transportedunit

4.2.1

Product-specific material

consumption (components)

++ lower amounts of materialsincluding proportional use of theserver

4.2.1

Use phase energy

consumption

++ best case: factor 4

worst case: factor 2-3

4.2.2

Recycling/Disposal + lower amounts of materials, fewercomponents

4.2.14.2.3

Space required on desk top +/0 depends on the arrangement ofthe work station (on top of orunderneath the desktop)

Requirement for additionalcomponents

4.2.2

Ergonomics/ Noise/ Wasteheat

+/0 any advantage depends on thework station (basic noise level)

Gas emissions were not examined

4.2.2

Benefit parity not comparable as system

performance dependent

on user profile

dependent on profile and on user'srequirements/preferences

4.1

++: significantly better +: better 0: the same -: worse --: considerably worse

Qualitativesummary of results


18th December 2006

2 Nature of the task

The subject of the examination is the environmental effects of a PC and a thinclient-supported supply of IT services. The manufacturing phase, the use phaseand the disposal phase are taken into consideration. For this a suitable ecologicalassessment instrument was selected (performance indicators, eco balance sheet,material intensity etc.) and modified for the client's specific requirements.Fraunhofer UMSICHT used the Umberto and GEMIS programs for the calculation.Here German values were used (e.g. for the power mix).

In addition the study included aspects of work ergonomics. These aspectsincluded comparable examinations of noise emissions and researching andgathering ergonomic findings on the effect of undesirable noise sources in offices.

3 Method

3.1 Definition of scenarios

In order to be able to compare PC and thin client desk top equipment, firstly"equivalent" PC and thin client systems must be identified. The systemrequirements differ depending on the user type and are described below1 . Thisentails developing the most realistic comparison scenarios possible. The clientsystems to be examined must be operated under these conditions.

As already explained in the economic examination "PC v. thin clients" (cf.[UMSICHT, 2006], P. 93ff), the question arose, primarily for server sizing2 , of how asystem must be equipped to be able to cope with a specific load successfully. Thisload is determined depending on the end user's workload. It has been establishedin the specialist literature that two to four sample user types including suitable usecases must be used.

Therefore, in the course of this study, as in the aforementioned economic study, amodel consisting of the three user groups, Light User, Medium User and HeavyUser was used again (see the following table).

Table 2: Different user groups

1 In order to assess energy efficiency this means that the relevant average power consumption in the operating phase must be takeninto consideration. A comparison of the nominal capacities of the respective power supply units is not meaningful. Here the specificenergy requirement in practical use must be assessed.

2 The term server sizing includes all matters regarding how a system must be sized in terms of processing capacity, main memory andprocessors in order to cope with the requirements of daily operation.


18th December 2006

User type Use Description

Light User Usually only uses one application atone time. Primarily a data recording ore-mail program is used.

Has only very low requirements forcomputing power and mainmemory.

Medium User Only uses two or three applications atthe same time. These include client/server applications with databaseaccess and also tools such as MS Office.

Computer power requirements aregreater than for a light user.

Heavy User Constantly uses several applications atthe same time, edits large graphics ordocuments, works a lot with "Outlook"and "Excel" for example, includingcreating diagrams, making calculationsusing large amounts of data.

Multi tasking operation, greatestamount of computer power requiredand large amount of main memoryrequired.

These user groups were allocated terminals depending on their requirements.This only included hardware that was directly connected with one or otheroperating model. Other infrastructure required, such as a file or mail server,which were needed regardless of whether PCs or thin clients were used, werenot taken into consideration.

3.1.1 PC systems

3.1.1.1 Light User

For the Light User, that is, for example, an administrator who works primarily withan application such as an ERP system client, a two year old desktop system wasselected, a system with moderate hardware specifications that can cope with therequirements of this user group. This takes into account the fact that older systemsare often used just for terminals that need few resources for a significantly longerperiod of time than is the case with terminals with higher requirements.


18th December 2006

Hardware specifications

Processor: AMD Athlon 1800+ RAM: 256 MB Hard disk: IDE (20 GB) Power supply unit: 300 W Housing design: Midi tower

3.1.1.2 Medium User

The Medium User , also known as a knowledge worker in specialist literature (cf.[Microsoft, 2003]) and runs several standard software products at the same time.In order to cope with this user type's requirements an up-to-date PC has beenincluded for testing such as the one currently used at Fraunhofer UMSICHT as astandard desktop.


Processor: Intel Pentium 4 (3 GHz) RAM: 512 MB Hard disk: IDE (80 GB) Power supply unit: 210 W Housing design: Mini tower

3.1.1.3 Heavy User

The Heavy User group works with not only standard applications but alsoresource-hungry applications such as image processing software. This user type isgiven a correspondingly more powerful computer.


Processor: Intel Pentium 4 (3.4 GHz) RAM: 1 GB Hard disk: S-ATA (160 GB) Power supply unit: 300 W Housing design: Midi tower

3.1.2 Thin Client systems

In the same way thin clients were subjected to examination depending on theuser category requirements. What all the thin clients have in common is thatinstead of a traditional hard disk they use a flash memory and thus, as both the


18th December 2006

processor and the power pack can run without active fans, there are no movingparts in the system.

3.1.2.1 Light User

An IGEL 2100 CE Smart thin client was selected as the terminal for the Light User.This is equipped with the current communication protocols, Microsoft RDP andCitrix ICA, which allow this user group access to any terminal server applications.


Processor: VIA Eden (400 MHz) RAM: 128 MB Hard disk: 128 MB (Compact Flash) Power supply unit: 20 W (external)

3.1.2.2 Medium User

On work stations for the Medium User , the "IGEL-3200 LX Compact" thin client isused. This is also used at Fraunhofer UMSICHT as a standard desktop terminal.Compared with the Smart series, the more powerful hardware offers the requiredperformance for effective working with more than one different terminal serversession.


Processor: VIA Eden (533 MHz) RAM: 128 MB Hard disk: 128 MB (Compact Flash) Power supply unit: 30 W

3.1.2.3 Heavy User

The "IGEL-5600 XP Premium" thin client was selected as the client for applicationswhich need a large amount of computer power. This allows the Heavy User toprocess data locally to a lesser extent using a local browser and other clientprograms, whilst native Windows applications continue to be run on the server.


18th December 2006


Processor: VIA C3 LP (1 GHz) RAM: 256 MB Hard disk: 512 MB (Compact Flash) Power supply unit: 48 W (external)

3.1.2.4 Terminal Server

In scenarios using thin clients, it is particularly necessary to bear in mind that theycannot be considered in isolation because they depend on applications run onthe server, which are provided for them by the terminal services. Thus, as thecomputer load is provided largely on the terminal servers, it is necessary to split theenergy requirements proportionately between the clients, which, in turn, raisesthe question of how many user sessions a single terminal server can run.

The subject of dimensioning and scaling of terminal servers has already beendealt with in a joint study carried out by Microsoft and HP. This study dividesgroups of end users into two categories, "data entry workers" who mainly dealwith entering data into a program and "knowledge workers", who work withseveral office applications in multi-tasking mode (cf. [Microsoft, 2003], page 9).Regardless of the fact that users today rarely only open one application butnormally run at least an e-mail client and a browser at the same time, the study isbased on the rather unrealistic values of only 3.5 MB of RAM for each data entryworker and 9.5 MB for each knowledge worker. Thus the evaluation found, forexample, with an HP ProLiant DL360 server, fitted with a maximum of two IntelXeon processors and 4,096 MB of RAM, depending on the category, between200 and 440 user sessions could be run at the same time. However, from arandom survey of the productive terminal server farm at Fraunhofer UMSICHT, itemerged that individual common application processes, such as Microsoft®

Outlook® or other Microsoft® Office Suite products each needed more than 20MB, which provides sufficient reason to examine the above study critically.

As part of the Fraunhofer study on the economic advantages of thin client workstations (cf. [UMSICHT, 2006], P. 93ff), an extensive number of preliminaryconsiderations on dimensioning and scaling terminal servers have already beenset out. During the study, starting from the question of how a specific load shouldactually be defined, hardware in the HP ProLiant DL360 performance class wasput forward as a recommendation. This decision was based on the followingconsiderations:

The critical components of an application server are the main memory andthe processor. Larger hard disk capacities are not necessary with the state of


18th December 2006

technological development today because data is kept on dedicatedmachines.

As standard systems with four or more processors normally include a complexmemory sub-system which is not relevant for the purpose of the application,these servers are disproportionately expensive. The most economical solutionis to buy dual processor systems to which further servers can be added whenthe performance limit is reached, working on the principle that “several small isbetter than a few large”.

On the assumption that individual user sessions each need at least 64MB RAMand the actual anticipated processor load can not be determined in advance,a maximum number of 35 Medium Users per dual processor was assumed.

Based on these preliminary points, in order to set up the productive terminalserver farm at Fraunhofer UMSICHT, the correct number of HP ProLiant DL360systems were purchased at the end of 2004 so that the above assumptions of thestudy could be verified after almost two years of productive operation. This ispossible because the required statistical data is available in the LANrunner®

system3 developed by Fraunhofer UMSICHT and proves that the informationprovided by the Microsoft®/HP study actually indicates values that are significantlytoo high, while, on the other hand, with an actual number of 35 users per server,the system is already working to full capacity. Monitoring a terminal server inFraunhofer UMSICHT's productive environment over 24 hours4 shows a clearcorrelation between the number of users and the working load of the system,where it is not the processors but the available physical main memory thatappears to be the limiting factor.

These were sessions with users who were connected to the terminal server for adesktop session and who were running several applications at the same time,such as the Microsoft® Office package, a web browser and database clients andthus came under the Medium User category.

3 Using the LANrunner® system network statistics can be calculated and displayed using the Simple Network Monitoring Protocol(SNMP) (http://www.lanrunner.de).

4 The 24 hour monitoring period was selected for clarity. Viewed over weeks and months it was verified that these measurements wereactually representative.


18th December 2006

Fig. 1: Active terminal server sessions over 24 hours

Fig. 2: Processor load for this terminal over 24 hours

Fig. 3: Server's available physical main memory over 24 hours

Based on the fact that Light Users each use individual applications which placefewer demands on system resources than the above products, it is expected thatan HP DL 360 server will be able to run up to 50 such sessions. This evaluation isnot only supported by our own measurement results but also by the experiences


18th December 2006

of Fischer Ges.m.b.H.5, which presented its terminal server environment as a casestudy at the Citrix iForum™2005 in Munich and, according to its own figures,provided more than 50 users per server with the client program for its ERP systemwith just the HP DL 360 server model.

On the other hand, the resources required by Heavy Users, who run a largenumber of applications at the same time in multi-tasking mode increasecorrespondingly and so it is assumed that the server can only cope with 20sessions of this type. In this study the same HP DL360 server hardware was usedto calculate the energy requirement for each of the three user types.

Server hardware specifications

Processor: 2x Intel XEON (3.4 GHz) RAM: 4 GB Hard disk: 2x 36.4 GB Power supply unit: 2x 460 W (1+1 redundant)

In order to take account of the different user types, the power consumption of thisserver was transferred to the clients based on the following factors:

Table 3: Allocation of the server's energy requirements to the clients (allocation key)

User type Transfer factor according to sessions per server

Light User 1/50

Medium User 1/35

Heavy User 1/20

3.2 Legal position

When manufacturing and marketing IT equipment, European regulations andtheir translation into national law must be taken into consideration. The followingregulations are particularly relevant with regard to environmental compatibility.

5 Fischer Ges.m.b.H. produces winter sports goods (http://www.fischer-nordicwalking.com/de/) and carbon parts for the automobileindustry (http://www.fischer-ct.com/de/fct/)


18th December 2006

Table 4: Legal requirements for computer systems (extract)

EU level Germany

Directive 2002/96/EC Wasteelectrical and electronic equipmentWEEE

Law dealing with the marketing, return andenvironmentally compatible disposal of electricaland electronic equipment (Electrical andElectronic Equipment Act – ElektroG)

Directive 2002/95/EC Hazardoussubstances in electrical andelectronic equipment (RoHS)

ElektroG

Directive 2006/12/EC on waste(replaces directive 75/442/EEC –EU outline waste directive)

KrW-/AbfG (Recycling management and wasteAct)

Under these directives, for example, a free collection system for private users for ITequipment to be disposed of is required and specific collection and recycling ratesare set out. Free return applies from 23rd March 2006. The recycling rate must beat least 75%, 65% of which must be for reuse or recycling of parts. In addition tothe requirements for the disposal phase, from 1st July 2006, specific material banshave come into force, (e.g. the ban on solder containing lead and the ban onsome flame retardants).

In all the following calculations the consequences and expected effects of thenew legal position are indicated.

Other European and national regulations for manufacturing and marketing ITequipment, such as safety-related regulations, are not examined in greater detailas part of this study.

3.3 System limits and material streams

The three phases of the product life cycle are looked at in this study - theproduction, use and disposal phases. In the production phase, various materialsand energy are used to build the IT equipment.

Only the direct effects of the individual phases were examined in this study. Theprevious chains, i.e. the extraction of metals and the production of fuels are notsummed up individually. After production, the desktop equipment is distributedand operated in the companies. During the use phase, the equipment requiresenergy (power) and creates emissions e.g. in the form of noise. After the usephase, the units are collected and disposed of or recycled. The disposal phase


18th December 2006

ends here. At IGEL Technology GmbH, systems are disposed of by Vfw AG6.However, hitherto there have been no findings and data on the return obligation[Reif, 2006]. Emissions caused by production are avoided by reusing part or all ofthe thin clients. Avoiding emissions must be included as a credit in the analysis.When computer systems are recycled, they are often first of all dismantledmanually before they are reduced to smaller pieces, sorted and then recycled infurther processes by machine.

Fig. 4: Manual dismantling of electronic equipment

Copyright: BMU/Rupert Oberhäuser

6 Further information can be found at : http://www.ctm.at/vfw/total.asp?peco=&Seite=259&Lg=1&Cy=1&UID=


18th December 2006

Fig. 5: Production, use and disposal phase abstraction

The individual phases are comprised of other processes such as various transport,recycling and production procedures. In addition, costs for packaging materialand their disposal must be borne in mind.

The energy and material costs for the individual phases and processes arecalculated. Offsetting procedures are used for credits (avoiding production andenvironmental pollution by secondary raw materials).

3.4 Selecting a suitable instrument for the ecological comparison of work station equipment

There are various assessment methods, including: EPS (Environmental PriorityStrategy), MIPS (Material intensity per service unit), the effect categories' method,e.g. of the Federal Environment Agency, the Centre for Milieu Customers' method,the calculation for energy-hungry products using CED (Cumulative EnergyDemand) etc. Compiling the life cycle inventory analyses, defining system limits,evaluating the effects on the environment and interpreting the results should beguided by the provisions of ISO 14040ff (Environmental performance evaluation).

As far as this study is concerned, environmental performance analyses are notapplicable because the life cycle inventory analyses required for these are not

Production phase

Energy

Primaryar-raw materials,,Secondaryraw materials

PC,Server,

Thin Client

Use phase

Disposal phase

Recycling,land fill


18th December 2006

available, particularly in the production and disposal phase and, because of thelow amount of data on material and energy consumption, cannot be compiledeasily. In addition, unfortunately environmental performance evaluations often donot lead to clear results.

As a result meaningful key figures to describe the individual phases were selectedand calculated for the study. The selected key figures create a basis forcomparison and simplify decision-making. Nevertheless, key figures must beinterpreted to reflect an unambiguous result.

It is important that they are as accurate as possible. Particular attention must bepaid if they do not measure the figure to be determined directly but work withindicators. Thus, for example, it is difficult to measure the effects on people of aPC's waste heat. However, the heat itself can be measured and used as anindicator for the calculation.

It is also important that one characteristic alone is not overvalued because it mustalways be assessed in conjunction with other relevant points. Thus the energysaving features of a PC should not be assessed just on the power consumption inoperation but also on the energy requirement for its production and disposalphase.

Individual characteristics do not always reflect the complex interrelationships. Thusit is not possible to ascertain precisely the required amount of raw materials for theproduction of a PC from one characteristic which indicates the amount of a PCthat can be recycled.

When using key figures, the traditional requirements for empirical research mustbe fundamentally met (objectivity, validity and reliability). This means, in principle,that they must meet the criteria below for creating key figures in accordance withthe KGSt (Municipal association for administration management) [KGSt, 1999]. Todo this, three target figures (control, economic advantage of collection andbenefits for the recipient) are used:

1. Control

- target -related- control-related- capable of being influenced- not susceptible to being manipulated

2. Economic advantage of collection

- availability- cost of collection


18th December 2006

- benefit for control- durability

3. Benefit for recipient

- comprehensible- unambiguous- quick to interpret- credible data basis

3.5 Selection of characteristics for comparing work station equipment

Possible characteristics can first of all be allocated to the product life cycle phaseswith these objectives.

It is important that, an individual PC should not be compared with an individualthin client because the thin client still needs a server.

The following energy and material characteristics will be used differentiated bythe production, use and disposal phase.

3.5.1 Production phase

The following must be borne in mind in production:

Transport processes: the comparison of the space requirement for thin clientsand the corresponding PC systems was made proportionately. Both pieces ofequipment are produced mainly in Asia, i.e. the transport distances andlogistics (distribution) are comparable. But the space required for the units instandard containers is important.

The amount of raw materials actually used in the end product (particularattention should be paid to toxic substances). Also of interest is the amount ofraw materials required to produce the materials which are added to theequipment at a later stage. One example of this is the production ofmicrochips which is very material intensive.

Amounts of waste (particular attention should be paid to toxic substances)

Energy consumed during production

Emissions produced during production, e.g. into the air. Emissions such aswaste heat, vapours produced by solvents etc., which arise during theproduction processes are not taken into consideration because a precisecalculation is not possible. Also the consumption of resources or the impacton resources such as water for cooling purposes is not to be calculated.


18th December 2006

3.5.2 Use phase

The main effects on the environment are produced during operation by, forexample, energy consumption,. PCs in the USA currently use up approximately 1%of national power consumption [Kühr 2003, P. 3]. The energy consumption pointis therefore more important during the use phase. Extensive comparisonmeasurements were carried out here as part of this study.

In this study, not only were the ecological effects examined for the use phase, butalso ergonomic aspects and, of course, the requirements of the regulations forworking with workstations [BildscharbV], that

when setting up the workstation the noise caused by the associated workingequipment must be taken into consideration, particularly in order to avoidproblems in concentration and the ability to understand other people, and

the equipment must not produce an unbearably high amount of heat at theworkstation.

In order to assess the exposure to noise, the thin client properties are compared interms of quality with those of the PC taking into account current ergonomicknowledge.

Other potential emissions of the equipment examined, such as the release ofhazardous substances into the surrounding air, electromagnetic fields, vibrationsand the associated effects on the temperature of the room and on people, werenot taken into consideration. Such an examination would have gone beyond thescope of this study.

In this study the space requirement was determined using the units' dimensions.

3.5.3 Disposal phase

In Germany alone, some 1.8 million tonnes of electrical scrap is produced eachyear. This includes 110,000 tonnes per year from the computing and IT field alone[BMU, 2006]. Therefore it is important to look at the number of PCs or thin clientsrecycled. Industry figures indicate that one in four brand new PCs goesimmediately from stores into recycling because it was developed too late or toomany units were produced [Radtke/Siegel, 2000]. This is another reason to payparticular attention to the disposal of equipment.

Another reason is environmental problems which occur when disposing ofelectrical waste, particularly in Asia. The organisation Greenpeace demands thatelectronics manufacturers supply longer lasting, non-toxic, easily recyclable,upgradeable, repairable products. On the other hand stricter controls on recyclingpractices in China and India are required, as well as measures against the illegal


18th December 2006

export of electrical scrap to Asia. Here, for example, people work under conditionsthat are extremely hazardous to health. Greenpeace has made these requestsbecause of investigations in New Delhi, India and Guiyu in China. There highamounts of hazardous materials were found on floors, in sediment, waste waterand ground water that were caused by the disposal and recycling of electricalscrap [Greenpeace, 2005].

In this study, we are striving for a breakdown by components and their recyclingrates. Therefore the following key indicators were established:

Amount of recycling: Here the recycled amounts (broken down by fractions,see above) were shown. This produces an environmental credit for theprocess because elsewhere production and thus the associated emissions ofthese parts may not apply.

Emissions, poisonous and hazardous materials that occur or are released inthe disposal process (this includes the sorting and separating processes andfinal removal of the fractions to the incineration plants or to landfill sites).

Collection rates: Here the collection system used was evaluated.

3.6 Data collection procedure

Data was collected from the literature to evaluate the individual phases in theproduct life cycle.

Through research in libraries and online databases, a selection of the literaturewas assembled. In addition data was obtained by questioning experts and fromstudies carried out by the Fraunhofer organisation. Further information on therelevant literature sources can be found in the sources' index.

This secondary data was also complemented and validated by our ownmeasurements in the laboratory.

3.6.1 Weight

As there was no precise information on the materials contained in the individualunits available from production, the units were dismantled into their individualparts if it was possible without destroying them. These parts were broken downinto electronic, plastic and metal parts and weighed with calibrated laboratoryscales.


18th December 2006

3.6.2 Power consumption

When measuring power consumption it must be borne in mind that the powerpacks in computer systems cause a phase shift between power and voltage as aresult of capacitive or inductive effects. The product of effective values of powerand voltage is the apparent power and is referred to as VA. The proportion ofapparent power that is actually "consumed", (i.e. used as power and finallyconverted into heat), is the active power and is referred to as W. The connectionbetween apparent and active power is characterised by the phase shift betweenpower and voltage. If the phase shift is zero, the apparent and active power arethe same.

Measuring equipment that does not take this phase shift into account records theapparent power but not the relevant active power for "power consumption".

In order to record the power consumption of the individual workstationequipment, it was connected to a Crompton Instruments' 16307 measuringinstrument. This measuring instrument indicates both the apparent power andthe active power. The latter is output as an integrated mean value over the courseof an hour beyond the current value. In addition the maximum mean value overthe whole measuring period is stored. The measuring equipment belongs to class0.2, i.e. it indicates the active power to an accuracy of 0.2% of the rangemaximum8. When the individual workstation systems had been equipped in thisway, measurements were carried out on them during Fraunhofer UMSICHT'sdaily routine while the staff corresponding to the light, medium and heavy useprofiles performed their normal work on the systems. In turn, the mean value wascalculated from the maximum mean values from five working days and used forfurther calculations within this study. In the same way the server was operated sothat only one of the two redundant 460 W power packs was connected to thepower supply.

In addition to the measured power consumption, the required cooling power forthe server that is normally housed in an air-conditioned room must be taken intoconsideration. Basically it must be assumed that 1 W of electrical powerconsumed by the server requires 1 W of thermal power for cooling. In order toproduce 1 W of thermal power, in turn, approximately 0.7 W of electrical powermust be used. In practice, however, the effectiveness of the air-conditioningdepends to a large extent on the air-conditioning equipment itself and localconditions such as the air flow in the room or external influences such as theamount of sunlight, so that more than 1 W of thermal power may be needed inorder to cool 1 W of electrical power. Therefore, in this study a ratio of the server's

7 http://www.crompton-instruments.com/german_downloads/de_catalogues/Integra_1630.pdf8 As a limiting factor it must be stated that this high accuracy is only achieved from 5% of the range maximum which led particularly to the

power consumption being beneath the trigger limit and therefore not taken into consideration when the units were in standby mode.


18th December 2006

active electrical power to the cooling system's active electrical power wasestimated as 1.

3.7 Comparison of workstation equipment, validation and sensitivity analyses

The key figures were ascertained for both systems and validated by the literatureresults. Then both systems were compared.

The main influencing factors on the system comparison are the user types (heavy,light and medium users) and the selected disposal methods for the equipment.Key elements here are the number of users that the server can cope with and thecredits for environmental pollution avoided. A third parameter was that the useperiods were varied.

The effect of these parameters on the results was calculated and interpreted.


18th December 2006

4 Summary and results from the use scenarios

4.1 Introducing the scenarios

As defined in section 3.1, the following systems were compared:

Table 5: Comparison of thin client and PC systems

Thin Clients comparedwith

PC work station system

IGEL-2100 CE Smart comparedwith

equivalent PC system

IGEL-3200 LX Compact comparedwith


IGEL-5600 XP Premium comparedwith


Proportional offset of expenses for aserver for each thin client

With the PC systems there areno server costs.

With the selected systems compared there is a parity of benefits for the user, i.e.he receives IT services (same computer power) of the same quality, regardless ofwhether he works with PC systems or with thin clients. Of course, services such asfile or mail servers run in the background with thin clients but this is also the casewith PC systems. Because of this there is the same complexity for both systems andthis does not contribute to more or less pollution of the environment.

In the literature it is possible to find values which give several years as the differentamount of time a PC and a thin client can be used [ETCF, 2006]. Basically thinclient systems can be used for longer than PC systems but in some circumstancessoftware innovation cycles provide for shorter periods. Many companies haveswitched to leasing workstation systems. Based on experience, an average periodof use of 4 years was used as a basis for PCs and thin clients.

This assumed an average use of 8 – 9 hours a day (220 working days a year) forthe thin clients and the PC systems. The server runs 24 hours a day.


18th December 2006

4.2 Results

The results are shown in the form of key figures for the production, use anddisposal phases and summarised clearly.

4.2.1 Production phase

Transport processes:

Transport weights (with transport packaging) and above all transport volumesplay an important role in transport processes after production. The following tableshows the data for the IGEL models examined and for the comparable systems:

Table 6: Transport weights and volumes

The weights of the IGEL models correspond to some 27-31 % of the weights ofthe comparable PC systems and only take up about 11-20 % of the transportvolume. When taking the server for the IGEL into consideration there is also aweight and volume advantage (here in the worst case with only 20 users perserver): The IGEL models only weigh 35% - 40% of the comparable PC systemsand only take up 18% - 30% of the volume.

Both pieces of equipment are produced mainly in Asia, i.e. the transport distancesand logistics (distribution) are comparable. But the amount of space required forthe units in containers is important here and an advantage for the thin clients isapparent. There is also a corresponding advantage for transport in the disposalphase.

Material consumption:

Normally with high-quality IT products, there is a very high product materialintensity, i.e. significantly more materials are transported and processed than arefound in the end products. In the literature it is assumed that the amount of rawmaterials used to manufacture a PC is equivalent to the amount used for an

Transport measurements IGEL-2100 Comparable -system

IGEL-3200 Comparablesystem


Server

Transport weight [kg] 2.6 8.5 3.0 11.0 4.0 12.9 16.0

Packaging size [dm³] 10.9 102.9 15.6 78.3 15.8 116.5 160.8


18th December 2006

average sports' car [Wolschk, 2006]. In another study, the amount of resourcesrequired to produce a PC with a monitor (CRT9) is estimated non-specifically as 22kilograms of chemicals, 240 kilograms of fossil fuels and 1,500 kilograms of waterused (see table 7) [Williams, 2003].

Table 7: Fossil fuels, chemicals, and water consumed in the production of one desktop computer

Item Fossil fuels(kg)

Chemicals(kg)

Water(kg)

Semiconductors 94 7,1 310

Printed circuit boards 14 14 780

CRT picture tube 9.5 0.49 450

Bulk materials – control unit 21 NI NI

Bulk materials – CRT 22 NI NI

Electronic materials/chemicals(excluding wafers)

64 NI NI

Silicon wafers 17 NI NI

Manufacture of parts NI NI NI

Assembly of computer NI NI NI

Total 240 22 1 500Notes: NI = not included in analysis. Only two significant digits have been kept in sums.[Williams, 2003]

The consumption of fossil fuels is calculated from the energy consumed duringproduction. In PCs, until the RoHs comes into force, problematic materials areused (solder containing lead, cadmium, flame retardants containing bromine andplasticisers in cables). These materials will occur in the disposal phase in future butthey are no longer included in new products. Metals for the main boards aresometimes transported over long distances from Africa and South America to Asiafor processing. In addition a great deal of energy is required for melting down theores. This is shown by the high energy intensity of the metals in the followingtable.

9 Cathode Ray Tube


18th December 2006

Table 8: Content and associated energy use in production of materials for desktop control unit

Material Energy intensityof material(MJ/kg)

Main use(s) in control unit Amount contained(grams)

Energycontent(MJ/unit)

Steel 59 Housing 6 050 360

Copper 94 Wires, circuit boards 670 63

Aluminium 214 HD, circuit boards 440 94

Plastics 84 Housing, CD-ROM 650 55

Epoxy 140 Circuit boards 1 040 150

Tin 230 Solder 47 11

Lead 54 Solder 27 1,5

Nickel 340 Disk drive 18 6.2

Silver 1 570 Circuit boards 1.4 2.3

Gold 84 000 Circuit boards 0.36 30

Subtotal 8 944 770

Other 96

Total 9 040 770

[Williams, 2003]

It is also apparent that a large number of materials have already been used in thepreliminary processing chains (reduction agents for iron manufacture or fuels fortransport). This material is no longer found in the end units. A full balance sheet ofthese material flows cannot be found in the literature. The reasons for this areconstantly changing computer models, the many different materials and thedifficulties of allocating these materials and the emissions associated with theprocessing procedures to the individual computer.

Therefore in this study we refer to the weight of components installed in the units.To do this, the equipment was dismantled by Fraunhofer UMSICHT as far aspossible without destroying it.


18th December 2006

Table 9: Weight of IT equipment components (all information in kg)

Here the weight advantages of IGEL products are apparent. This advantageremains even when apportioning the material installed in the server. Here theworst case scenario with only 20 users per server was examined.

Table 10: Weight of IT equipment components, offsetting the server (all information in kg)

No conclusion can be drawn about the hazardous nature of the materialsinstalled in the components. It may certainly be the case that small amounts of atoxic material have a serious harmful effect on the environment. But this pointcould not be included in this study.

How easily a product can be recycled and its energy efficiency are increasinglyimportant. The key phrase is Design for Environment (DfE). A specification sheetfor the IGEL 5600 shows that in an ideal case 99.8% of the materials used can berecycled. Normally about 76% of the materials are recycled (which is significantlymore than the 65% rate legally required in Europe). The recycling rate ofapproximately 83% is also far above the 75% required.

Amounts of waste in production:

At this time there was no data available from IGEL Technology GmbH. This data isdifficult to allocate to the computers and in addition different legal conditions from

Material characteristics [kg] IGEL-2100 Comparablesystem



Server

Electronics, (circuit board,processor, main memory,,expansion cards etc.)

0.4 0.9 0.5 1.2 0.9 1.2 2.5

Plastic, (front cover,base etc.)

0.1 0.4 0.1 0.3 0.2 0.4 0.4

Metal (housing,slot face plates etc.)

1.1 3.9 1.4 5.9 1.7 7.2 6.1

Power pack (2 power packsfor server))

0.2 1.5 0.2 2.0 0.5 2.0 2.3

Material characteristics (afterapportioning the servermaterial to theIGEL) Worst case: 20 users per

server in [kg]



IGEL-5600 Comparable-system

Server

Electronics (circuit board,processor, main memory,expansion cards etc.)

0.5 0.9 0.6 1.2 1.0 1.2 2.5

Plastic (front cover,base etc.)

0.1 0.4 0.1 0.3 0.2 0.4 0.4

Metal (housing,slot covers etc.)

1.4 3.9 1.7 5.9 2.0 7.2 6.1

Power pack (with server2 power packs)

0.3 1.5 0.3 2.0 0.6 2.0 2.3


18th December 2006

those in Germany apply in the production sites. One way of collecting data forthis (and also to cover potential savings), would be to compile a waste balancesheet for the production sites.

Energy consumed during production:

At this time there was also no data available from IGEL Technology GmbH. Buthere too economic advantages could be found with savings.

The "Environmental Consciousness" study by the American professionalassociation, MCC (Microelectronics and Computer Technology Corporation) hasinvestigated how to produce PCs more cheaply and in a more environmentally-friendly way and has provided figures for manufacturing (shell and core of acomputer including monitor approximately 3,000 kWh. Adding to this costs fordismantling and transporting the raw materials gives approximately 5,300 kWh foran average PC) [Hoffmann, 2004].

Emissions

No clear conclusions are possible on this issue because of the lack of data. Data isonly available for separate components eg. circuit boards.

The following table shows general values for manufacturing a PC. But this datamust not simply be transferred to IGEL production.

Table 11: Emissions when manufacturing a PC with monitor

Emissions For 1 PC For 2 million PCs

Hydrocarbons 0.01 kg 20 t

Carbon monoxide 0.02 kg 40 t

Dust 0.01 kg 20 t

Nitrogen oxides 1.25 kg 2,500 t

Sulphur dioxide 2.14 kg 4,280 t

Carbon dioxide 1,850 kg 3,700,000 t

As a result air polluted to the limit 1 million m³

Waste 60 kg 120,000 t

[Radtke/Siegel, 2000]

Air pollution is only part of the problem. According to IBM, this becomes clearbecause, in order to process a 20g silicon wafer for memory chips or processors,28kg. of liquid chemicals are required in more than 400 work stages, which thenmust be neutralised by using a further 11kg of chemicals. Every day some 70,000


18th December 2006

m³ of highly pure nitrogen is required for a state of the art production site forsemi-conductors. In order to manufacture a PC and monitor, around 33,000 litresof cooling water are required [Radtke/Siegel, 2000].

4.2.2 Use phase

Energy consumption in the use phase

As the thin client has to supply significantly fewer components with energy thanthe traditional PC, it must be assumed that it consumes significantly less power.This assumption has been checked with a large number of measurements.

The advantage of thin clients can also be seen with energy consumption. Thenext table shows the consumption including the proportionate allocation of theenergy required by the server. Of course, the voltage peaks sometimes reachedsignificantly higher levels in normal operation, e.g. with graphic-intensiverequirements such as playing back a film or a flash animation. However, thesepeaks only occurred for a limited time and were under the mean value. Here thinclients only consume a third of the power of a PC.

Table 12: Comparison of energy consumption in the use phase

Even when including the cooling power for the server, which has been estimatedconservatively as double the required power, thin clients use significantly lessenergy than PCs (factor 2) – see next table.

Table 13: Comparison of power consumption in the use phase including the cooling power for the server

Energy requirement IGEL-2100 Comparablesystem



Server

Power consumptionAverage active power [W]

14 90 16 68 19 96 246

Power consumptionActive power with server pro rata(worst case: 20 users / server)[W]

26 28 31 246

Energy requirement IGEL-2100 Comparablesystem



Server

Power consumptionAverage active power [W]

14 90 16 68 19 96 246

Power consumptionActive power with server pro rata+ server cooling, (double theenergy requirement with worstcase: 20 users/ server) [W]

39 41 44 246


18th December 2006

Recently, energy consumption and the emissions associated with this have beenenjoying a high level of political interest with the trading of emiss ion rights etc. Aspart of this study, emissions per kWh consumed were calculated with the GEMISand UMBERTO programs.The production of one kWh with the German electricity network gives rise to thefollowing emissions: 0.63 kg CO2, 0.0001 kg CO, 0.00002 kg PM10

10, 0.0009 kgNOx etc. In the following table emissions into water and metal emissions into theair have been left out to make it clearer.

Table 14: Emissions when producing one kWh in the German electricity network

Table 14: Emissions in the production of one kWh in the German electricity network

Balance Sheet Input/OutputInput Output

Item Quantity Unit Item Quantity UnitCumulative energydemand (CED)

Waste…

CED (nuclear energy) 4.14E+03 kJ Waste for removal(WfR)

CED (water power) 6.87E+00 kJ Excavated material(WfR)

3.02E+00 Kg

CED (fossil fuel – total) 6.49E+03 kJ Special waste (WfR) 2.90E-06 KgCED other, regenerative 1.24E+02 kJ Waste for recycling

(WfRec)Raw materials indeposits (RiD)

Filter dust (WfR) 2.30E-03 Kg

Energy sources (RiD) Gypsum (RED)(WfRec)

7.95E-03 Kg

Natural Gas (RiD) 2.19E-02 Kg Coarse ashes(WfRec)

3.44E-04 Kg

Oil (RiD) 1.50E-03 Kg Sodium sulphate(WfRec)

7.56E-05 Kg

Coal (RiD) Melting chambergranulate (WfRec)

2.89E-03 Kg

Brown coal (RiD) 3.43-01 Kg Fluid bed ash(WfRec)

2.30E-04 Kg

Hard/black coal (RiD) 9.07E-02 Kg Emissions (air)Uranium (RiD) 7.08E-06 Kg Waste heat (A) 6.11E+03 kJNon energy sources Dust (>PM10) (A) 9.37E-06 Kg

10 PM10 = Particulate matter < 10 µm


18th December 2006

(RiD)Minerals & ores (RiD) Dust (PM10) (A) 2.19E-05 KgLimestone (RiD) 1.36E-02 Kg Compounds,

anorganic (A)Sodium chloride (RiD) 7.66-05 Kg Ammonia (A) 5.69E-06 KgSand (RiD) 2.69E-04 Kg Hydrogen chloride

(A)2.04E-05 Kg

Raw materials, basic Dinitrogenmonoxide (A)

8.06E-06 Kg

Sulphur (RiD) 5.00E-05 Kg Fluorine (A) 9.28E-11 KgWater Hydrogen fluoride 2.89E-06 KgProcessing/Drinkingwater

Carbon Dioxide (A)

Water (coolant) 2.33E+02 Kg Carbon dioxide,fossil (A)

6.35E-01 Kg

Water (boiler feed) 5.89E+01 Kg Carbon monoxide(A)

1.45E-04 Kg

Water (process) 1.02E-01 Kg Metals (A)Water (unspecified) 4.50E-03 kg NoX (A) 9.05E-04 Kg

Radio-nuclides (A)Radio-nuclides, total(A)

4.32E+05 Bq

Sulphur (A) 6.50E-10 KgSulphur dioxide (A) 4.91E-04 KgHydrogen sulphide(A)

6.45E-10 Kg

VOC (A)Methane (A) 1.24E-03 KgNMVOC (A)NMVOC, halog. (A)NMVOC, chlor. (A)NMVOC, chlor.aromat. (A)Chlorobenzene (A) 7.42E-17 KgChlorophenol (A) 1.48E-16 KgPCB (A) 7.42E-19 KgPCDD, PCDF (A) 1.28E-14 KgNMVOC, fluor (A)Perfluorothane (A) 2.60E-11 KgNMVOC, non-halog. (A)Aldehyde (A)Formaldehyde (A) 3.27E-08 KgAlkanes (A)Hexane (A) 1.18E-07 KgAromaticcompounds (A)Aromatichydrocarbons (A)


18th December 2006

Benzole (A) 3.21E-08 KgPAC (A)Benzo(a)pyrene (A) 2.33E-11 KgPAC without B(a)P(A)

6.93E-09 Kg

NMVOC, unspec.(A)

2.06E-05 Kg

Emissions (water)Energy sources(secondary)Energy (electrical) 3.60E+03 kJArea of unspoiltnatureLand fill volumes 4.29E-06 M**3Nuclear waste,highly radioactive,ultimate wastedisposal

4.20E-09 M**3

Nuclear waste,medium level ofradioactivity,ultimate wastedisposal

1.17E-09 M**3

Nuclear waste, lowlevel ofradioactivity,ultimate wastedisposal

6.37E-09 M**3

WaterWaste waterWaste water(clarified by boilerfilter presses)

5.98E+01 Kg

Waste water(coolant)

8.41E+01 Kg

Waste water(process)

1.10E-02 Kg

Waste water(clarified)

9.81E-04 Kg

Seepage, diffuse 1.16E-04 KgSeepage, collected 2.36E-05 KgWater vapour 1.49E+02 Kg

kJ 1.08E+04 kJ kBq 3.01E+03 kBqKg 2.92E+02 Kg kJ 1.11E+03 kJ

Kg 2.96E+02 KgM**3 4.30E-06 M**3


18th December 2006

[Umberto 2006]

The Eco Institute's GEMIS program gives similar results [GEMIS, 2006]. Here CO2

emissions are 0.61 kg, for CO 0.0002 kg, for NOX 0.0006 and for dust 0.00005 kg.Dust emissions are higher because all the dust was calculated and not only thedust particles smaller than 10 µm.

The active power consumed by workstation equipment is ultimately released inthe form of heat and thus contributes to increasing the temperature of thesurrounding area. Because of the low amount of heat they emit compared withPCs, thin clients have fundamental advantages when used on workstationswhere particular measures are required to avoid a harmful heat load. As the heatload at the workstation depends on a large number of other factors (e.g. heating,ventilation, air-conditioning, direct sunlight) no differentiated statements on thisaspect can be made in this study.

Exposure to noise at the workstation

Exposure to noise at the workstation has a negative effect on the concentration,performance and productivity of the staff working at the workstation. If it isparticularly necessary to maintain high levels of concentration for long periods atthese workstations, according to ergonomic knowledge, troublesome pollutantsfrom the surrounding area should be avoided if possible [Probst, 2003a; 2003b].

Binding upper acoustic pressure assessment level limits as set out in workplaceregulations and German trade association's regulation BGV B 3, are used toprevent employees being damaged by excessive noise with the focus on avoidingdamaging hearing. At 85 dB(A) the critical limit for this is far beyond the ratinglevel acceptable for office activities. In the old version of the workplace regulationswhich ceased to be valid in August 2004, a limit of 55 dB(A) was set for jobsinvolving predominantly mental activities.

Such limits do not take sufficient account of more recent ergonomic knowledgeon the existing requirements for workstations. As a result of ergonomicinvestigations [van den Brulle, 1995; Probst, 2003] significantly lower target valuesfor workstations in offices have been proposed. Consequently rating levels of upto 30 dB(A) are ideal, up to 40 dB(A) very good, up to 45 dB(A) good and up to50 dB(A) acceptable in a commercial environment. These target values relate to allnoises that are not caused by the person in question himself and therefore areperceived as disruptive when work requires concentration.

In single use offices without significant external sources affecting them, a level ofless than 30 dB(A) is normal; in open plan offices ideally a range of between 40and 50 dB(A) is reached [Probst, 2003a].


18th December 2006

An important criterion for the disruptive effect of a noise is not only the A assessedacoustic pressure level (noise level for short) of the noise averaged over time butthe identifiability and information content. A persistent noise such as the whistlingof a fan, which stands out clearly from background noise has a considerably moredisruptive effect than a neutral noise with the same noise level that cannot beattached to a specific source. According to Probst [Probst, 2003a] a workstation inan office environment where high levels of concentration are required must beclassified as "harmful" if the noise from an individual sound source exceeds theproportionate noise level of the other sources taken altogether by more than 4dB(A).

Because of the aforementioned correlations, the noise caused by a PC (fans, harddrives) also has a disruptive effect and leads to the workstation being assessed as"harmful in terms of noise" if the noise level produced just on its own would leadto a "good" rating. This applies particularly in very quiet single offices.

Because of the way they are constructed with no fans or hard disks, thin clients donot create any of the background noises typical of PCs such as fans rustling orwhistling or hard disk noises. Also thin clients do not cause any deterioration insound emission behaviour while they are being used. Therefore thin clients areparticularly suitable where it is critical to avoid the disruptive effects of noise inwork places which require a high level of concentration for a long time.

The noise emitted by the server (when using a thin client) is not significant fromthe point of view of exposure to noise at the workstation, if it is assumed that theserver is set up in a separate room where there are no permanent workstations.

When assessing the noise made by a computer it must be taken intoconsideration that when working at the workstation other sources of noise,sometimes with significantly higher acoustic power arise. Thus Probst [Probst,2003a] specifies an acoustic power of 45 dB(A) as a planning value for the soundemission of a PC when idle (fan, hard disk) . The acoustic power level of using thekeyboard is already 60 dB(A), that of a person speaking on the telephone 65dB(A) (planning values). If these noises are not caused by the person in questionhimself, (e.g. in an open plan office), these noises must also be considereddisruptive [Probst, 2003a]. The effect of a reduction in computer noise that can beachieved by using a thin client instead of a PC puts this into perspective.

Space required on desk top

The amount of space required was calculated using the dimensions of theequipment. Here the amount needed for the server is not offset proportionatelybecause it is not housed in the office space.

Table 15: Comparing the space requirement


18th December 2006

By taking up less space than a PC, the thin client offers ergonomic advantages inthe workplace. The saving in space creates advantages for users particularly innarrow, densely occupied offices.

But here too reservations must be made when for some users the PC can beplaced under the desk but typically the thin client is placed upright on top of thedesk.

4.2.3 Disposal phase

Quantities recycled

A PC consists of different components which provide a challenge for recycling.The table below shows a summary of the main components of a PC.

Table 16: Main components of a PC

Module/ component Environment-related materials

Fitted circuit board Bromine, cadmium, quicksilver, nickel, lead, tin, zinc, silver,aluminium, gold, copper, iron, glass, plastics

Condensers etc. PCB, aluminium, iron, plastics

Batteries/Accumulators Cadmium, lithium, nickel

Housing Bromine, cadmium, chlorine, nickel, lead, tin, iron, plastics, steel

Cable Chlorine, copper, PVC

Peripheral equipment Aluminium

Cathode ray tubes Aluminium, barium, cadmium, lead, strontium, magnesium, glass

[Kitow, 2006]

In Table 9 it was shown that thin clients contain significantly fewer componentsand thus weigh less than PC systems. Because of the restricted amount ofhardware, small size and low weight, they thus create significantly less waste thana comparable PC.

Reusing or recycling the equipment have the greatest benefits for theenvironment. Currently only just under 10% of a computer is recycled. However,this proportion could be significantly higher if all manufacturers consistentlyemployed used parts for repairs.

Ergonomics IGEL-2100 Comparablesystem



Server

Space requirement in dm³ 1.5 31.7 2.2 27.1 3.5 37.4 15.0Footprint in dm² 0.1 0.8 0.1 0.7 0.1 1.0 3.4


18th December 2006

According to the manufacturers, a great deal is recycled - but large parts of a PCcannot be recycled because they are not of a sufficiently high value. Between 5and 10% of total volume must be permanently disposed of. Normally the plasticcomponents of a PC are difficult to recycle because they only appear as acomposite with other materials and can not be identified. As composites, they canno longer be processed into high-quality material.

As far as batteries are concerned, nickel cadmium batteries are highly toxic butcan be recycled relatively well. Nickel metal hydride or lithium ion batteries are notnormally so harmful to the environment but the recycling system is not yet readyfor them [Radtke/Siegel, 2000].

Unfortunately the disposal procedures for PC systems and IGEL systems cannot bebalanced precisely. The reason for this is the recycling and return systems whichwere introduced only this year. Thus credits for recycled components cannot beincluded.

The effect of recycling rates on the system comparison is explored in greater detailin the sensitivity analysis. This deals with the proportions of materials.

Emissions, toxic and hazardous materials

Preparing IT equipment incurs costs for energy, staff and if necessary, othermaterials. This includes staff for the sorting processes, energy costs for breakingthe equipment down into smaller pieces and further separation of material. Someof the materials will end up in land fill sites and in waste incineration plants. Asthese preparation processes have never been quantified before no more preciseconclusion was possible. Credits that arise from recycling material would have tobe included again.

Collection systems

IGEL Technology GmbH offers a free equipment return service for customers.Because of this, high collection rates are possible which offers an advantage overother chargeable return systems for business.

4.3 Summary of the results (comparison of workstation equipment)

In comparing the equipment, the advantages of thin clients were apparentwithout exception:

Thin clients have a weight advantage in transport weights and packaging sizes.They are only 35-40 % of the weight of a PC and only take up 19-30% of thevolume. An overseas container therefore holds considerably more thin clientsthan PCs.


18th December 2006

It can be assumed that there are advantages with material consumption becausethin clients are made up of fewer components (circuit board, processor, passivecooling unit, chip, memory chips for the operating system or main memory, cable,housing).

The advantage of thin clients is also found in energy consumption. Consumptionis at least twice as low, sometimes even three or four times lower that theconsumption of corresponding PC systems. This applies even with theproportionate offsetting of the energy required by the server and the coolingpower required for this.

Because of the way they are constructed with no fans or hard disks, thin clients donot create any of the background noises typical of PCs such as fans rustling orwhistling or hard disk noises. Also thin clients do not cause any deterioration insound emission behaviour while they are being used. Therefore thin clients areparticularly suitable where it is critical to avoid the disruptive effects of noise inwork places that require a high level of concentration for a long time.

In the use phase, the thin client saves space for the user particularly in narrow anddensely occupied offices.

The key material figures were determined by weighing the individual parts. Theyshow the environmental friendliness of the equipment in the disposal or recyclingprocess. The more recyclable parts there are the more environmentally-friendlythe product's classification. At the same time a low total amount of materials isdesirable in order to reduce the overall consumption of materials. The IGEL 5600data sheet shows that up to 99% can be recycled and that the normal recyclingrate is approximately 76%.


18th December 2006

4.4 Validation and sensitivity analyses

Validation

For this study, our own measurements and above all the book "Computer and theEnvironment" were the main data sources. The results correspond with the studiesand analyses researched.

Sensitivity analyses

Three main influential factors, whose influence is to be investigated below, weredetermined in section 3.7.

The first factor is the user types. This factor specifies the number of clients perserver and thus indirectly the whole system's energy consumption and materialrequirement. A worst case assumption is 21 heavy users because here the twenty-first user makes an additional server necessary. But the results still show a lowerconsumption for thin clients compared with the PC solutions. With the server, therequired cooling power was included.

Table 17: Power consumption sensitivity – user types (worst case)

In a realistic scenario with 40 medium users and 10 heavy and 10 light users, thinclients are approximately twice as good as far as energy consumption isconcerned.

Table 18: Power consumption sensitivity – user types (normal case)

IGEL PC1 641 2 016

Total powerconsumption [W]

Number ofserversrequired for IGEL

Total number ofunits

2.0021

IGEL PC2 407 4 580

Total powerconsumption [W]

Number ofserversrequired for IGEL


2.0060


18th December 2006

As far as material consumption is concerned, thin clients in the worst casescenario (21 heavy users) are two and a half times better than the PC solutions.

Table 19: Material consumption sensitivity – user types (worst case)

In a normal case (10 heavy users, 10 light users and 40 medium users), thin clientsare three times better for material consumption.

Table 20: Material consumption sensitivity – user types (normal case)

The second influential factor examined is the use period. Material consumptionwas calculated per year. To do this the thin clients were compared with therelevant PC models (after offsetting the proportional material costs for the server).The advantages of thin clients can be clearly seen with a use period of 4 years forthin clients and the PC systems: They are about three times better for materialconsumption. As the thin client theoretically has a longer service life than a PC, inreality the results for the thin client are even better.

IGEL PC90.3 226.6


Number ofservers required

Total materialconsumption [kg]

21 2for IGEL

IGEL PC161.5 548.1


Number ofservers requiredfor IGEL

Total materialconsumption [kg]

60 2


18th December 2006

Table 21: Material consumption sensitivity per use year with a use period of four years

The third factor is the recycling rate. On the assumption that some of the materialsare reused in the equipment, the material requirement for new units is lower.With a 65% recycling rate only 35% of materials, for example, would have to beobtained from primary raw materials. Thin clients normally offer good possibilitiesfor recycling. Here too the PCs are not significantly better and so thin clients stillhave the advantage.

5 Interpretation of the results

The status quo was examined in this study. The market is in a constant state of fluxbecause of environmental requirements etc. so some key figures (e.g. for energyconsumption) may change. Here the energy efficiency for electrical equipmentdebates should be followed (e.g. the EU Directive on secondary energy efficiencyand on energy services) and the Design for Environment (DfE) debates in which itwill be necessary to produce electrical equipment so that it can be recycled.Additional requirements may arise in the future because of developments on aninternational scale because the international community is currently discussingworld wide minimum standards for dealing with electrical scrap in view of themost recent environmental catastrophes [BMU-Presse, 2006].

In comparing the systems in this study, it is assumed that the benefits of thin clientsand PCs were identical. This applies to normal standard office applications (Office

"Material requirement"[kg]

per year for one unit




HP DL360(Server)

Electronics (circuit board,processor, main memory,expansion cards etc.)

0.11 0.21 0.15 0.30 0.25 0.31 0.62

Plastic (front cover,base etc.)

0.03 0.10 0.03 0.08 0.05 0.10 0.11

Metal (housing,slot facing plates etc.)

0.31 0.99 0.39 1.47 0.50 1.80 1.53

Power pack (with server2 power packs)

0.07 0.37 0.06 0.49 0.15 0.50 0.58

Total material weight in[kg/a]

0.52 1.66 0.63 2.34 0.95 2.70 2.83


18th December 2006

etc.). In comparing the equipment the advantages of the thin clients wereapparent without exception:

Thin clients have a weight advantage in transport weights and packaging sizes.They are only 35-40% of the weight of a PC and only take up 19-30% of thevolume. An overseas container therefore holds considerably more thin clientsthan PCs.

It can be assumed that there are advantages with material consumption becausethin clients are made up of few components (circuit board, processor, passivecooling unit, chip, memory chips for the operating system or main memory, cable,housing).

With energy consumption, it is possible to see the advantage of thin clients.Consumption is at least twice as low, sometimes even three or four times lowerthat the consumption of corresponding PC systems. This applies even with theproportionate offsetting of the energy required by the server and the coolingpower required for this. Thus between 27 and 54 W are saved. The other ITequipment should be taken into consideration in order to be able to classify thisvalue: A 17" TFT monitor consumes about 20 W. A comparable 19" cathode raymonitor about 85 W. This alone is a difference of 65 W.

Thin clients are better in terms of sound emissions because of the way they aremade. Therefore thin clients are particularly suitable where it is critical to avoid thedisruptive effects of noise in work places that require a high level of concentrationfor a long time.

In the use phase, the thin client saves space for the user particularly in narrow anddensely occupied offices.

Other possibilities for use arise in the industrial environment where thin clients candemonstrate a lower breakdown rate and a longer service life than PCs becausethe fan-free units are less susceptible to malfunctions, particularly in dustyenvironments.

The key material figures were determined by weighing the individual parts. Theyshow the environmental-friendliness of the equipment in the disposal or recyclingprocess. The more recyclable parts there are the more environmentally-friendlythe product's classification. At the same time a low total amount of materials isdesirable in order to reduce the overall consumption of materials. The IGEL 5600data sheet shows that up to 99% can be recycled and that the normal recyclingrate is approximately 76%.

Sensitivity analyses relating to use periods, user types and recycling rates confirmthe results.


18th December 2006


18th December 2006

6 Sources

[B]

[BAuA, 2006] Bundesanstalt für Arbeitsschutz und Arbeitsmedizin (BAuA):Akustische Gestaltung von Bildschirmarbeit splätzen in Büros.Reihe » Technik«, Broschüre Nr. 26, 4. Aufl., Dortmund 2006

[BildscharbV] Verordnung über Sicherheit und Gesundheitsschutz bei der Arbeit anBildschirmgeräten, BildscharbV - Bildschirmarbeitsverordnung vom4. Dezember 1996, BGBl. I S. 1841

[BMU, 2006] Bundesministerium für Umwelt- Naturschutz und Reaktorsicherheit:Elektroschrott – Vermeiden und Verwerten.http://www.bmu.de/abfallwirtschaft/elektro_und_elektronikgeraetegesetz/doc/36726.php

[BMU-Presse, 2006] Bundesministerium für Umwelt- Naturschutz und Reaktorsicherheit:Pressedienst Nr. 322/06: »Basler Konvention/Internationales« vom 4.Dezember 2006

[E]

[ETCF, 2006] European Thin Client Forum, Isartalstr. 49, 80469 München;www.etcf.de/cms/branche_benefits_de/

[ElektroG, 2006] Gesetz über das Inverkehrbringen, die Rücknahme und dieumweltverträgliche Entsorgung von Elektro- und Elektronikgeräten(Elektro- und Elektronikgerätegesetz – ElektroG);http://www.bmu.de/files/pdfs/allgemein/application/pdf/elektrog.pdf

[G]

[GEMIS, 2006] Öko-Institut: Globales Emissions-Modell Integrierter Systeme; Prozess: Netz-el -DE-lokal-HH/KV-2000

[Greenpeace, 2005] Untersuchungsbericht von Greenpeace über Elektrorecycling in Asien;http://www.greenpeace.de/elektroschrott

[H]

[Hoffmann, 2004] Hoffmann, Ausarbeitung in der Informatik bei Dipl.–Informatiker Märtensan der FH Wolfenbüttel http://public.rz.fh-wolfenbuettel.de/~maertenh/InfGes_WS0405/Ausarbeitungen-/Hoffmann.pdf

[I]

[IGEL, 2006] IGEL Technology GmbH: Unter die »grüne« Lupe genommen – ThinClients versus Arbeitsplatz-PChttp://www.igel.de/ps/tools/download.php?file=/templates/lmcontent/psfle/download2_de/38/WP_Green_d445f67e78226e.pdf&name=BG_Green_de_web.pdf&id=1746&nodeid=250&_language=de

[K]


18th December 2006

[Kitow, 2006] Zusammenfassung über die Umweltbelastungen von PC von Svenja Kitow,IGEL Technologie, Marketing

[KGSt, 1999] Kommunale Gemeinschaftsstelle für Verwaltungsmanagement (KGSt);Ziele finden, Zahlen kennen, Handeln können;http://www.kgst.de/menu_oben/iko_netz/aktuelles_aus_dem_iko_netz/neuer_flyer_iko_netz/index.html

[Krems, 2006] Online-Verwaltungslexikon; Dr. Burkhardt Krems, Professor an derFachhochschule des Bundes für öffentliche Verwaltung,Fachbereich Allgemeine Innere Verwaltung;http://www.olev.de/k/kennz.htm

[Kühr 2003] Kühr, Rüdiger; Williams, Eric (Hrsg.): Computer and the Environment –Understanding and Managing their Impacts; Dorddrecht et al:Kluwer, 2003; ISBN: 1-4020-1680-8

[M]

[Microsoft, 2003] Microsoft Corporation (Hrsg.), 2003. »Windows Server 2003 TerminalServer Capacity and Scaling«;http://www.microsoft.com/windowsserver2003/techinfo/overview/tsscaling.mspx (10/2003)

[Microsoft, 2005] Microsoft Corporation (Hrsg.), 2005. »Terminal Services Scaling andPerformance on x64-Based Versions of Windows Server 2003«<https://www.microsoft.com/downloads/details.aspx?familyid=9B1A8518-D693-4BBB-9AF8-B91BBC0D2D55&displaylang=en>(12/2005)

[P]

[Probst, 2003a] Probst, W.: Beurteilung und Minderung des Lärms anBildschirmarbeitsplätzen im Büro und in der Produktion.Arbeitswissenschaftliche Erkenntnisse Nr. 123: Bildschirmarbeit –Lärmminderung in kleinen Büros. Bundesanstalt für Arbeitsschutzund Arbeitsmedizin (Hrsg.), Dortmund, 2003

[Probst, 2003b] Probst, W.: Beurteilung und Minderung des Lärms anBildschirmarbeitsplätzen im Büro und in der Produktion.Arbeitswissenschaftliche Erkenntnisse Nr. 124: Bildschirmarbeit –Lärmminderung in Mehrpersonenbüros. Bundesanstalt fürArbeitsschutz und Arbeitsmedizin (Hrsg.), Dortmund, 2003

[R]

[Radtke/Siegel, 2000] Ökobilanz eines PC, Ausarbeitung in der Informatik bei Prof. Dr. Junker ander freien Universität Berlin

Zitiert nach: Hage, M.; Beschonrner, T.; Kuhn, J.: Nachhaltigkeit imBedürfnisfeld Information und Kommunikation – Ergebnisse einerExpertenbefragung; Universität Oldenburg; Institut für ökologischeWirtschaftsforschung iöw; April 2004; GELENA-DiskussionspapierNr. 04-03;http://www.ioew.de/home/downloaddateien/gelena_wp_04_03.pdf

[Reif, 2006] Experteninterview mit Kilian Reif von der Vfw AG vom 13. November 2006


18th December 2006

[T]

[Tritsch, 2003] Tritsch, Dr. Bernhard, Microsoft Windows Server 2003 Terminaldienste,Unterschleißheim: Microsoft Press® Deutschland, 2003

[U]

[UMBERTO, 2006] Umberto 4.3; ifu Hamburg GmbH, Institut für Energie- & UmweltforschungHeidelberg; Modul Strom Deutschland

[UMSICHT, 2006] Fraunhofer-UMSICHT (Hrsg.), 2006. »Wirtschaftlichkeitsbetrachtung - PC vs.Thin Client«<http://cc-asp.fraunhofer.de/docs/PCvsTC-de.pdf> (04/2006)

[V]

[van den Brulle, 1995] van den Brulle, P.: Schalltechnische Gestaltung von Büroräumen mitBildschirmen, Schriftenreihe Forschungsanwendung Fb 720,Bundesanstalt für Arbeitsschutz und Arbeitsmedizin, Dortmund1995

[VBG, 2000] Leitfaden für die Gestaltung von Bildschirm- und Büroarbeitsplätzen vonder Verwaltungsberufsgenossenschaft (VBG)http://www.eso.org/safety/Archive/VBG_Brochure_Bildschirm_SP21.pdf

[W]

[Williams, 2003] Eric Williams: »Environmental impacts in the production of personalcomputers«, in: Kuehr, R.; Williams, E. (ed.): Computers and theEnvironment: Understanding and Managing Their Impacts,Dordrecht/NL, October 2003

[Wolschk, 2006] Pressemitteilung des European Thin Client Forum (ETCF) vom 04.09.2006

Documents

Study - Axess Systems · Ecological comparison of PC and thin client desk top equipment compiled for: IGEL Technology GmbH Mr. Heiko Gloge Schlachte 39/40 28195 Bremen / Germany Oberhausen,