• No results found

We would like to direct our regards to Fresenius Kabi AB for letting us con-

N/A
N/A
Protected

Academic year: 2021

Share "We would like to direct our regards to Fresenius Kabi AB for letting us con-"

Copied!
59
0
0

Loading.... (view fulltext now)

Full text

(1)

Foreword

We would like to direct our regards to Fresenius Kabi AB for letting us con-

duct our Thesis work within their organization. A special thanks goes to

our tutor Joakim Bohlin at Fresenius, for excellent guidance and consulta-

tion regarding both big and small matters. We would also like to thank our

supervisor Erik Hultman at Uppsala university, for continuous support re-

garding project planning and execution, as well as for the structure of the

report. Furthermore, we would like to thank Martin Nilsson at Pivac AB,

for sharing your expertise and knowledge.

(2)

Populärvetenskaplig sammanfattning

Läkemedelsbranschen där Fresenius Kabi är verksamma är strikt kontrollerad för att säkerställa produkternas kvalitet. Detta medför stor försiktighet vad gäller arbetssätt och genomförandet av förändringar vid fabriken. Även det minsta fel kan få förödande konsekvenser.

Att automatisera materialflöden inom fabriken är en sådan förändring som företaget önskar genomföra. Detta tros ge bättre arbetsförhållanden för de anställda samt bidra till ökad driftsäkerhet. En del av denna framtida au- tomation är hanteringen av påsar innehållande 3500 caps som används till behållare för intravenös näringslösning. Detta arbete syftar till att bevisa ett

"proof of concept" för att på ett säkert och repeterbart sätt lyfta en sådan påse, öppna den samt hälla ut innehållet.

Den presenterade lösningen i detta projekt innefattar en 6-axald industrirobot som är utrustad med ett gripverktyg. Verktyget består av en ledad alumini- umarm som, via pneumatisk styrning, utför en nypande rörelse om påsens ena kortsida. Åtta sugkoppar fixerar påsen i verktyget vilket möjliggör fri hantering av påsen. Robotcellen är utrustad med en knivstation där en kniv är monterad på en pneumatiskt opererad linjärenhet. Kniven skär upp den ena sidan av påsen varpå roboten tömmer innehållet i ett tråg.

Arbetet har lett fram till konstruerade prototyper som med goda resultat

testats både enskilt och med en 6-axlad industrirobot. Genom robotsimuler-

ing och ekonomiska beräkningar bedöms enbart robotcellen inte ge frabriken

någon ekonomisk vinning. Genom att även automatisera materialflöde för

cellen uppskattas hela automationen vara återbetald inom loppet av två år.

(3)

Contents

1 Abbrevations and definitions 5

2 Introduction 6

2.1 Background . . . . 6

2.2 Ongoing changes at the factory . . . . 8

2.2.1 Automation of material flow . . . . 8

2.2.2 Assignment . . . . 9

2.3 Responsibilities within the project . . . . 10

2.4 Project boundaries . . . . 11

3 Theory 12 3.1 GMP and FDA regulations . . . . 12

3.1.1 Clean room . . . . 12

3.2 Vacuum technologies . . . . 12

3.2.1 Vacuum ejectors . . . . 13

3.2.2 Suction cups . . . . 14

3.3 Plastic cutting . . . . 16

3.4 Industrial robot . . . . 16

3.4.1 Hardware . . . . 17

3.4.2 Robot programming . . . . 18

3.5 Investment calculations . . . . 18

4 Method 19 4.1 Prestudy . . . . 19

4.2 Ideating . . . . 19

4.3 Prototyping . . . . 20

4.4 Testing . . . . 21

5 Designs and process 22 5.1 Manufacturing parts . . . . 22

5.2 Pinching the bag . . . . 23

(4)

5.3 Gripping the bag with vacuum . . . . 25

5.4 Cutting station . . . . 29

5.5 Robot tool assembly . . . . 30

5.6 Robot simulation . . . . 31

5.7 Economy . . . . 33

6 Result 34 6.1 Final designs . . . . 34

6.2 Economical scenarios . . . . 36

6.3 Robot simulation . . . . 36

7 Discussion 37 7.1 Is automation the way to go? . . . . 37

7.2 Vacuum switch . . . . 38

7.3 Gripping the bag with vacuum . . . . 38

7.4 Choice of suction cups . . . . 39

7.5 Equipping the robot cell with a PLC . . . . 39

7.6 Manufacturing parts . . . . 40

7.7 Cutting station . . . . 40

7.8 Pinching tool . . . . 41

7.9 Repeatability . . . . 41

7.10 Flexibility of the robot cell . . . . 42

7.11 Regulations of a clean room . . . . 42

7.12 Progress towards an automated material flow . . . . 43

7.13 Robot simulation . . . . 44

7.14 Safety considerations . . . . 45

7.15 Economical calculations . . . . 45

8 Conclusion 46

9 Recommendations and further development 47

10 Bibliography 48

A Appendix 51

B Appendix 57

(5)

Chapter 1

Abbrevations and definitions

Cycle The time it takes for the robot to perform one iteration of the robot program.

IV bag Intravenous nutrition bag AGV Automatic Guided Vehicle

GMP Good manufacturing practice, a standard that the pharma- ceutical industry among others must follow.

FDA Food and Drug Administration. The control organ that approves, for instance, drug products for the American market.

Clean Room A restricted area which is only allowed to contain a certain number of particles of different sizes per cubic meter air.

There is also a restriction to how many living microorgan- isms the air may contain.

PLC Programmable Logic Controller.

SRS Stäubli Robotics Suite, software for programming Stäubli robots.

Jogging Freely working with a robot’s movement control, which en- able physical or simulated movements in one or more axis without code writing.

VAL 3 Programming language used in Stäubli Robotics Suite.

HMI Human-Machine Interface. Interactions between system controls and operator

SolidWorks Software for 3D modeling.

(6)

Chapter 2 Introduction

2.1 Background

The strive of getting factories more efficient by automating different tasks has been going on ever since the 18th century. This was the time of the first industrial revolution where steam powered machinery was the

groundbreaking technology [20].

Since then, technology has become more advanced and precise. Today we are on the verge of the so called "fourth industrial revolution". Electric robots are more common than ever in factories around the world, many tools and machines are connected in a network and artificial intelligence enables the machines to handle more and more complex tasks. In many occasions robots are a necessity to keep up with competition and customer demands [17, 27].

In some countries, like Sweden among others, where the electricity is cheap and the labour is relatively expensive, robots and automation is a key factor to maintain a competitive and strong industry [23]. Automation might also be a solution to a problem that has emerged the latest decades.

That is how to manage the decreasing amount of population within working age in many countries. By automation, it is possible to get a single

operator to manage the work of multiple employees working manually [10].

Although this might also be seen as an issue and a threat. If the machines

becomes too efficient and intelligent, will there be any need for manual

labour what so ever? This has been a fear for every industrial revolution

and so far it has shown to be unfounded [3].

(7)

It is widely known that companies often experience several advantages with automation. The human factor is eliminated, improved workflow, increased production rate, less wear on the employees bodies and often the

traceability improves since a machine doesn’t misplace products or forgets to document certain things. But there are also drawbacks, it’s expensive to invest in automation. There might be a negative impact on the social aspect among employees on workplaces. Changes in the production may be very difficult to handle [15].

In Fresenius Kabi AB’s factory for parenteral nutrition in Uppsala, many tasks are already automated. However, one moment in the production line that is still done manually is refilling the caps for the intravenous nutrition bag (IV bag). Figure 2.1 shows an example of an IV bag with two models of caps mounted in the bottom.

Figure 2.1: IV bag with caps mounted

In every production line there are six different slots containing caps and

there are three different types of caps. Hence there are two slots for every

type of cap. To keep the production going, there always have to be caps in

every slot. The operators working in the production line have to perform

different tasks during their shifts. This includes taking samples of the

number of particles in the room, supplying the machine with plastic

materials to make the IV bags, making sure all the batch numbers match

and keep an eye out for how many caps there are in each slot to know when

to fill it up.

(8)

When the caps in a slot is beginning to run out, the operator needs to get a bag from a trolley, scan the bag and write down it’s batch number, open the bag with a pair of scissors, lift it up above shoulder height and pour the content into the slot. However, the slot that is being used today is not big enough to contain all the caps from a single bag. The operator needs to fill the slot to about 80 percent of what it may contain in order to avoid the caps jamming in the slot outlet. The bag with the remaining caps then needs to be set aside so that the next refill will come from the same bag.

This task means high and heavy lifts for the personnel. Continuously monitoring how many caps there are in the slot and how many bags there are in the room so that it can be refilled at the right time. There is also a risk of loosing the traceability needed. Every bag with caps needs to be scanned and the batch number needs to be matched to the batch number of the nutrition fluid. The opened bag needs to be stored so that everybody working in the production line knows which bag is currently being used.

2.2 Ongoing changes at the factory

At Fresenius Kabi’s factory the strive for continuous development is high.

New ideas of how to further increase production capacity, product quality and working environment are constantly investigated. One idea is of how an automation of material flow to and within the clean rooms can be achieved.

2.2.1 Automation of material flow

With today’s way of working as a foundation, an automated solution for

material flow is sought after. An automated way of handling, distributing

and sorting material throughout the factory is thought to improve both

production quality and the working environment. An idea of how to

implement this is to have one or two robot cells that could handle the

material and sort it into cartridges that are stored in a buffer. This buffer

would be equipped with a sensor system that could tell which cartridges are

full and which are empty. A number of AGVs (Automatic Guided Vehicle)

equipped with a smaller robot, see Figure 2.2, would then be able to collect

full cartridges and distribute them to the production line that currently

lacks material. The AGV would then return to the robot cell and insert the

empty cartridge into the buffer station. The design of a robot cell that could

handle a bag of material and distribute the content is therefore desirable.

(9)

Figure 2.2: Example sketch of an AGV for handling cartridges to and from a buffer station [28].

2.2.2 Assignment

This project is about a proof of concept in how to automate a certain moment in an existing production line. It can be seen as a part of the larger project described above. This specific task is to grab a 75x48 cm plastic bag containing plastic caps for IV bags, seen in Figure 2.3. The total weight of the bag is about 5,5 kg. The bag is to be cut open, the content should be poured into a tray and the empty bag is to be disposed in a suitable way. All this should be done with a robot and a specially designed gripping tool. An industrial robot, of model Stäubli RX160L, was provided to the cause of the project.

Figure 2.3: Working object, bag containing plastic caps.

(10)

The project has its foundation in the main question:

1. Is it possible to design and construct an automated robot cell, which is capable of opening a type of plastic bag and distribute the content?

Within this main issue, several partial questions are treated:

2. Which regulations needs to be taken into concern when implementing the robot cell in a C-classed clean room? Which adaptations must be made when choosing method of opening the bag and material for a robot tool?

3. How could the robot tool and associated equipment be constructed in order to meet requirements from question 1 and 2?

4. Could an automation of the today manually performed labour be profitable?

5. In which ways could the automation affect the working environment, production capacity and traceability?

The task should, in an as large extent as possible, be modeled so that it can be performed in a clean room area in a factory. Hence it must be

investigated how the bag should be cut open without leaving an excessive amount of particles behind. Further it should be taken into consideration how this can be done without risking damage or injury to goods and personnel. The economical situation regarding profitability of an

automation are to be compared with today’s working method. A simulation of the robot cell will be set up to give an estimate of cycle time.

2.3 Responsibilities within the project

The project is divided into sections where the two participants are responsible for one half each. Johannes Malmström is responsible for finding a suitable way of opening the bags (e.g. cutting technique), model the design of the final robot cell and assemble a test arena based on this design. Furthermore, he is responsible for economical calculations and conducting tests in the test arena. Vincent Sollie is responsible for

designing and constructing a robot tool which is compatible with a 6-axis

robotic arm. The tool should be able to pick up and handle the bag in a

desirable way. Vincent is also responsible for programming the robot and

(11)

set up a simulation to test cycle time.

2.4 Project boundaries

To be able accomplish this project within the amount of effort and time the course intends, there have to be strict limitations to what is included in the project.

The first limitation is to set a fixed starting position of the bag. This way the robot will always know where the bag is initially placed and that it’s not folded in any way. How the bag ends up in this position is not taken into consideration in this project. Another limitation is that there is no strict time limit for a cycle. The main target of this project is not to find the optimal way of implementing this in a factory, but rather to find one solution on how it may be done.

The area of the robot cell will not be limited by anything other than the reach of the robot. To scan and register the batch numbers of the bags will not be covered in this project. The tray that the caps will be poured into will be of arbitrary shape and size and will be able to contain all caps from a bag at once. The material flow to and from the robot cell will not be taken into consideration. The materials used to produce the parts of the tool and robot cell is only suitable for the draft and not, unless otherwise specified, recommended to use in any further implementations.

Safety recommendations will only be brought up for discussion in the

report, not specifically implemented in any drafts or models produced

during this project.

(12)

Chapter 3 Theory

3.1 GMP and FDA regulations

One of many demands on the pharmaceutical industry in Sweden is that they must comply to the GMP (Good Manufacturing Practice), which aims to minimize the risks of errors to manufactured products that affect the human body, like medicines or food. An often used complement to GMP is the FDA (Food and Drug Administration). These two standards are very much alike in many ways but to enable the American market for the product in question it’s important to follow both.[1]

3.1.1 Clean room

A clean room is an area with restrictions to the amount of particles that is allowed to occur. This entails that it needs to have a controlled air flow to, from and within the area. There are different standards to clean rooms. At Fresenius’ factory in Uppsala the standard of EU’s GMP applies. The robot cell, when and if implemented, should be located in clean room class C according to this standard. This translates to ISO 7 in standby and ISO 8 during operation[24].

3.2 Vacuum technologies

The definition of a volume beeing exposed to a complete vacuum is when

there exists no matter within the volume. However, vacuum is commonly

referred to as when the pressure within the arbitrary volume is noticeably

lower than that of the surrounding atmosphere[12]. In practice this means

(13)

that if the dynamic pressure within, for instance a suction cup, is achieved to be lower than outside the cup, vacuum is considered to be established.

3.2.1 Vacuum ejectors

While there exists a variety of ways of generating vacuum, one way is by using a so called vacuum ejector. The ejector is using the Venturi principle, which says that when a gas is traveling through a narrowing it’s flow rate and dynamic pressure increases[19]. Figure 3.1 shows a vacuum ejector which is mounted directly on top of a suction cup. Compressed air is

pumped through port A and fed through a nozzle inside the ejector, port B.

This generates a vacuum in D and by extension the suction cup. Port C functions as an outlet for the gas injected at A and sucked in through D [26]. In this figure, an external silencer is mounted on the outlet.

Figure 3.1: A multi stage vacuum ejector with suction cup and silencer.

When choosing the appropriate vacuum ejector for an application some

factors should be considered. Factors like a dirty environment, suction

force, cycle time and material of the object are decisive for choosing a

suitable model [2]. One difference between vacuum ejectors is the concept

of mono versus multi stage. The mono stage ejector is on its simplest form

with a single nozzle. This delivers a powerful vacuum once a sufficient air

flow is attained. Because of its simplicity, the mono stage ejector is built to

(14)

either supply a high vacuum flow or a high vacuum level. This is however not an issue for the multi stage ejector, as it in general is able to generate high levels of flow and vacuum at the same time.

The multi stage vacuum ejector is based upon the same principle but with several narrowings, which causes the ejector to react in multiple steps. In practice, this means that the working range of the ejector is wider and it is capable of generating vacuum at lower levels of compressed air. This may result in a quicker evacuation time of the suction cup, since the ejector starts to generate vacuum sooner. The multi stage ejector is often equipped with membranes in the nozzles which also serves as air filters. This is especially useful when working in a clean room. Any contaminations, in the compressed air or from the working object, are then filtered out and

contained within the ejector.

When determining which model of ejector to choose, one have to look into the demands of each application. A mono stage ejector is in general cheaper in purchase but consumes more compressed air. Hence, the mono stage model might be preferred for a time limited application. A work object that, due to it’s inconvenient surface or density, generates a high air leakage in the suction cup might require an ejector with focus on air flow in order to maintain vacuum. When working with a stiff and flat but heavy object, a proper level of vacuum is often prioritized [21, 5].

3.2.2 Suction cups

In robotics, it’s common to use suction cups as a gripping tool. They can lift objects of different shape, size, weight and material. Suction cups are also gentle to the workpiece and relatively simplistic in it’s construction.

There are several factors to take in consideration when choosing what suction cups to use in different applications. First of all it’s good to calculate an approximate size and theoretical holding force. This can be used to search through different data sheets to get an assumption on what diameter[11] and how many cups needed to handle the workpiece safely[26].

The shape of the workpiece have a great impact on what kind of suction

cups should be used as well. If it is a flat or slightly bent surface and the

material is rigid, flat cups without bellows might be a good choice. If the

shape of the workpiece is round or irregular one might choose a cup with

several bellows to ensure a better abutment. Although, time is often of

(15)

essence in production lines and it is desirable to make the lifts and

transports as fast as possible. This is a reason to keep the inner volume of the suction cups as low as possible. More bellows means bigger inner volume and more air to evacuate, so it’s not wanted to have too many bellows where it’s not necessary[29].

If the workpiece is long and narrow an oval suction cup might be the best option and if the workpiece is of a soft material i.e. a plastic bag, the suction cup might need to have internal ribs on the flap to keep the material from collapsing. There are applications where the workpieces are dirty or oily. In these occations it’s important to use a cup that can handle these conditions well. Other parameters to keep in mind is if the cup is going to be exposed to chemicals, extreme temperatures or if the components being handled is sensitive to static electricity.

The approximate size of the cups is evaluated in Equation 3.1 [11] and the theoretical holding force is calculated by applying Equation 3.2 [26]

D = 11, 2 ∗ p

(m ∗ s)/(b ∗ c) (3.1)

where

D = Diameter in mm m = Mass in kg.

c = Number of cups b = Vacuum in bar s = Safety factor

The safety factor should be at least 2 for horizontal lifts and at least 4 for vertical or tilting lifts. This is to compensate for different variables in the surroundings.

F

T H

= F

s

∗ n (3.2)

where

F

T H

= Theoretical holding force

F

s

= Suction force (from data sheet)

n = Number of suction cups

(16)

3.3 Plastic cutting

There are already several solutions on automation applications that picks up and cuts plastic bags. However, none of these are especially helpful in this occasion. In every one of the found solutions, the material inside the bag was not taken into consideration. If one of these systems were to be implemented into this project, most certainly several caps would be

damaged [6, 25, 8]. This is highly undesired. However, some different ways of lifting bags with care was found, but these systems could not provide a good cutting position [9, 7].

Since the bag is to be cut open in clean room class C (see subsection 3.1.1), it is desired that the cut leaves as few particles as possible behind. To be certain what method is the best way to achieve this, clinical studies in an environment representing the clean room needs to be done. This is very expensive and time consuming to perform. A good estimate on what is a suitable way to open the bag can be obtained by observing present techniques to open the bag in the production line. Then combining these observations with experiments to see if the bag is cut open cleanly or torn open [4].

Variables that can alter throughout the experiments are type of knives, angle of the knife, speed of the cut, position of the bag a.s.o. A specific technique is the so called tomato cut [4], where the point of the blade pierces the bag in a pushing motion before the cut is made.

3.4 Industrial robot

While the term robot can vary vastly depending on who is using it, in many applications concerning industrial automation a more correct term is

industrial robot. According to the international ISO standard, an industrial robot is an "automatically controlled, reprogrammable, multipurpose manipulator, programmable in three or more axes, which can be either fixed in place or mobile for use in industrial automation applications" [13].

Regarding industrial robots, flexibility is a commonly discussed term.

Whether a robotic solution should have a high flexibility or not is a choice

that needs to be made based on each application’s requirements. A robot

with a higher flexibility can in general handle different kind of objects,

(17)

perform multiple tasks and is less sensitive for changes in production rate etcetera [16].

3.4.1 Hardware

A frequently used family of the industrial robots in today’s automated industry are the six-axis articulated robots. These robots typically use three axes for positioning and three for orientation. The axis configuration enables the robot to access a point in space from any direction (within its reach) [30]. Each axis is controlled by a separate servo motor which makes it possible for the user to freely operate the robot in three dimensions [22].

A typical robot cell includes a few elements that works together. Figure 3.2 shows a robot tool (5) that is oriented and positioned by the manipulator (1), which is controlled by the control system (2). An option for manually jogging the robot is to use a controller (3). The robot program is otherwise configured using a PC with installed software (4). The figure also includes an example of a working object (6).

Figure 3.2: Example of hardware setup in a robot cell.

The user needs to be able to program and communicate with the robot.

One way of doing this is by using the robot’s associated control system.

While processing the current application, the integrated control system may also be used for sending and receiving data signals. Such signals may

contain information regarding sensors, current working cycle, robot tool

controls or instructions from the HMI (Human Machine interface) [5].

(18)

Another way is to implement a PLC (Programmable Logic Controller), which serves as a hub between the host computer, manipulator and the robot’s control system. The advantages a PLC may bring are that the PLC lifts the "intelligence" from the control system [22]. This is often favoured when it comes to handling a larger count of signals or when several robots are to collaborate. The same PLC is also compatible with a multitude of robot control systems from different manufacturers, which eliminates the need to stick to one brand [5].

3.4.2 Robot programming

Stäubli Robotics Suite (hereinafter SRS), is a software for programming industrial robots from the manufacturer Stäubli. However, depending on the manufacturer of the actual robot, different softwares are available. SRS offers a graphical environment for jogging and teaching the manipulator as well as writing code in VAL 3 language.

3.5 Investment calculations

An investment is always more or less expensive and should be evaluated against its expected change in cash flow. One way of estimating an

investment’s profitability is to use the payback method. It gives a figure on how long it takes for an investment to be profitable according to

Equation 3.3 [14]:

t = Cost/C

f

(3.3)

where,

t - the payback time in years

Cost - the cost of the investment and

C

f

- expected positive change in annual cash flow, generated by the investment.

This is interesting because the longer it takes for a investment to be

profitable, the bigger risk it is for the investor.

(19)

Chapter 4 Method

4.1 Prestudy

The first days of the project were spent on site at the factory. This provided information about how the work is done today, what the difficulties are and how often the caps needs to be refilled. Experts in different areas within Fresenius Kabi’s organization were asked for input and advise regarding economy, clean rooms, automation and the present working structure.

The next step was to get some inspiration and to research if something similar had been done before. This was done by using large search engines on the web and the Uppsala university’s online library.

It was decided that the project would be performed in an agile way. This implies that the work is to be done in short sprints towards minor

milestones. Along with this a time plan and working structure was set. To make sure the project was kept within a reasonable workload, strict

boundaries were set. These are viewed in section 2.4.

To get a good estimate of the cost of the investment, offers on hardware were requested from suppliers. Experts within the company were consulted for cost of labour as it is done today and previous experience on costs for similar projects. Hence a payback time could be calculated.

4.2 Ideating

To come up with different solutions on how to pick up the bag with a robot

and cut it, an ideating session took place for the next few days. Aiming to

(20)

find a solution that hopefully would work and was realistic to achieve with the circumstances given. This session mainly consisted of brainstorming and discussing ideas about what is possible to do within the time frame of the project. Discussions occurred mostly between participants of the project, but also with supervisors and innovation experts.

The main goal of the project was never the direct subject for these

discussions, rather how minor tasks were to be achieved. These minor tasks were stated as questions much like "How might we":

• "pick up the bag?"

• "cut the bag?"

• "ensure that the caps don’t suffer any damage?"

• "pour the content in a controlled way?"

• "dispose the bag?"

• "build the tools?"

• "set up the robot cell?"

4.3 Prototyping

In parallell with the later bit of the ideating phase, prototypes were being produced. To get a clear view of the ideas on how the tools should work and look like, the first prototypes were drawn in SolidWorks. The sketches were rough and not too detailed, with the main purpose to show functions.

This enabled a more directed discussion and unnecessary misunderstandings could be avoided.

Often each of the participants drew their own idea by themselves, presented the designs to each other and discussed which parts of each design were the best. When an agreed design had been evaluated, the prototype was

produced in simple manor with materials that could be found on site or at a regular hardware store, i.e. plywood and clamps. The prototype was used in simple tests and the results of these tests were evaluated to see which parts of the design that could work and which that had to be redesigned.

Some ideas were even thrown away and that part of the project went back

into the ideating phase.

(21)

After some iterations a design that was worth moving on with had been found. Now a search for suitable materials and components began.

Studying data sheets to find out characteristics like weight, strength, air consumption a.s.o. along with testing and consulting experts and salesmen.

4.4 Testing

Throughout the project, testing have been a big part of finding the right solutions. Many ideas have been put to the test at the workshop on Fresenius technology center in Uppsala. By having access to vacuum, compressed air, 3D printers and a whole mechanical workshop it have been possible to assemble and test components individually. If the test result wasn’t acceptable, adjustments and changes to the components were made on site to enhance its’ performance. The component could even be replaced with a similar product with a bit different characteristics to observe if it was a better match to the application in question. When the individual designs were functioning properly, the components were assembled and tested together.

An area which would represent a robot cell was set up as soon as all the final designs were approved. This enabled testing of a robot cell without using a robot just to see if the whole task could be performed as desired.

The next step included the robot tool being mounted on the industrial

robot, in order to test the tool and associated equipment with movements

from a robot. This was done by teaching the robot specific robot targets

and jogging the robot between them. Another test was performed in SRS

where the thought robot cell was simulated in order to determine an

approximate cycle time and to generate the basics of the robot program.

(22)

Chapter 5

Designs and process

5.1 Manufacturing parts

The ideating phase initially resulted in different designs drawn in 3D and a prototype made from wood and clamps. Some of these can be seen in Appendix B. After testing and discussion, it stood clear that a suitable way to grip and open the bag was with a hinged clamp and a stand-alone knife, both operated by compressed air and a vacuum controlled gripper.

Components like hinges and mounts for the clamp and the knife were drawn in SolidWorks and initially printed in plastic with a 3D printer. The base of the robot tool was made out of aluminum profiles with built in rails.

Tested separately, the designs worked as intended. When put together and operated by the compressed air, the plastic parts showed to be too fragile.

Figure 5.1: Vacuum regulators used to trim behavioural properties of pneu-

matic actuators.

(23)

To get around this problem, the parts were manufactured in aluminum and the effects of the compressed air was controlled by regulators seen in

Figure 5.1. This resulted in smoother motions, longer lasting parts and safer handling.

5.2 Pinching the bag

The fist test was to lift the bag by hand to see how it shaped, when put from a flat position to an up-right position, depending on gripping points of the lift. After this the same tests were done but by fixating the bag only with clamps on a few points on the edge of the bag.

It showed that the upper seam above the plastic weld was a suitable gripping area. Gripping the bag with clamps in the upper edge still gave a good, up-right handling position, but the few gripping points of the clamps made the bag hang uneven and become nodular.

(a) Tool with two linear actuators

(b) Tool with one inbuilt linear actuator

Figure 5.2: Two designs of the clamping tool

Two different tools that clamped down the whole length of the bag’s upper

edge were designed. One spring loaded, working along two steel rods and

one long clamp that was hinged on the middle, creating a triangular shaped

opening when pushed. The first designs of these tools created in

(24)

SolidWorks can be seen in Figure 5.2.

After testing, a spring loaded tool was discarded quite soon because of difficulties getting it to work smoothly without getting stuck. The long hinged clamp proved to be more fail-safe. A piston driven by compressed air showed to be a simple and well functioning way to make the clamp open and close.

To create a first prototype of the hinged moving part of the clamp, two aluminum profiles were cut to the same length. A standard hinge was bought and mounted between the two profiles so that it became one longer jointed part. One end of the part was jointed by a self manufactured hinge and the other end was fixed to the piston via the connecting rod. The complete first draft can be seen in Figure 5.3. A second design made out of an aluminum bar was cut and milled to get an integrated joint, this

minimized the play and gave a more robust construction. The second design was mounted the same way as the first prototype. In Figure 5.4 the final design of the clamp is shown and in Appendix A there are pictures of the jointed parts.

Figure 5.3: First prototype of the clamp

Figure 5.4: Final design of the clamp

A piston, driven by compressed air, was integrated into the fixed part of the

clamp. This piston operated the moving part of the clamp via a connecting

(25)

rod. The piston itself could be operated by a manual switch.

When tested, the first prototype of the clamp proved to be too imprecise and could not perform as desired. The other design resulted in a more stable and precise clamp. By further testing it was discovered that the clamp could not be opened properly. To aid the opening movement of the clamp, an actuator, seen to the left in Figure 5.4 had to initiate the movement by pushing the jointed part in the right direction.

In order for the bag to stay in it’s position, different materials were attached to the gripping surface of the clamp, simply to get more friction between the clamp and the bag. First a rubber belt was cut in half and glued to the surface. Later the belt was replaced by high friction

polyurethane. To enhance the friction even further and to make sure the bag was kept in place, even during robot movements, the jointed part of the clamping tool was provided with spikes. These spikes were placed in such a way that when the clamp was closed, the spikes pinched the bag in a slot between two strings of yellow polyurethane.

5.3 Gripping the bag with vacuum

Since the bag had different materials on its front and back, one with

transparent plastic and one with woven plastic, the first tests were designed to try out suction capability on the two. It showed that the woven, non transparent side, proved to be inappropriate for this task since it was found to leak too much air. The transparent plastic gave a significantly better suction force due to reduced leakage of air through the material.

Further tests were performed in order to find a satisfying method of vacuum generation, including vacuum ejectors, a vacuum cleaner and generating vacuum by only suppressing the air inside a suction cup by hand. It quickly became clear that in order to achieve a usable level of vacuum, an ejector was the best choice. The ejector provided an easy way of turning the vacuum on and off and the security of having multiple vacuum generators, which the vacuum cleaner did not. The ejector also excelled by its minimal size. Compared with only compressing the suction cup by hand, the ejector contributed with a steadier level of vacuum.

Tests were performed to find a sufficient suction cup. Initially, self made

(26)

suction cups were constructed using plastic foam, which were cut into different shapes. Figure 5.5 shows three different models tested on the bag of caps. The tests were performed with a consistent vacuum flow for easier comparison.

Model 5.5a in Figure 5.5 comes with six larger cavities functioning as vacuum chambers. It was almost able to lift the bag from the underlying support but was sensitive for vibrations. The lips of the model leaked air, resulting in lost vacuum. Model 5.5b differs from the other two by not having any distinct lips but instead a larger set of cups. This model reacted quickly when applied to the bag but did not reach a sufficient vacuum level.

The model tended to mostly grip the bag with the center holes, only leaving the outer ones to leak air. The third model, 5.5c, comes with a single wider lip. The flexible lip made it possible for the cup to adapt to the instability of the bag, almost providing sufficient vacuum to lift the bag.

However, the three models presented in Figure 5.2 shared a problem with maintaining vacuum during a non horizontal lift. A slight rotational movement resulted in air leakage and a lost vacuum.

(a) (b) (c)

Figure 5.5: Three designs of suction cups made from plastic foam

In search of a suction cup solution with better performance, tests of

different suction cups were performed during a field study at Pivac AB in

Täby, Sweden. Several cups of different models were tested with a bag of

caps in order to find the most suitable equipment. At the same field study,

tests for finding the right model of ejectors were performed. Conducted

tests included mono stage inline ejectors and multi stage models which were

mounted directly on the suction cup.

(27)

Tests at Pivac AB showed that a suction cup with a set of bellows and a thin and highly flexible lip was suitable for the application. The thin lip made it possible for the suction cup to adjust its shape to that of the bag, as seen in Figure 5.6. This showed to be important as the shape of the bag changed during lifts and depended on the acceleration of the lift. Since the bag was not perfectly flat when lying down but slightly convex, using several bellows proved to be favorably. This way, the bellows themselves were able to compensate for eventual level differences and angles that followed from the convexity of the bag.

Figure 5.6: Suction cup on bag of caps showing how flexible lips adapt to the surface of the working object

The inline ejectors were found to lead to a more flexible assembly as they do not have to be mounted directly onto the suction cup but only in adjacent to it. When comparing vacuum flow, tests showed that the multi stage ejector excelled over the inline model. Based on underlying theory and conducted tests, a multi stage type of ejector was chosen. In order to provide a high safety factor, in case one or more ejectors should fail, all suctions cups were equipped with one ejector each.

At a later state of the project, further testing took place in order to determine optimal placement of the suction cups and ejectors. It became clear that the suction cups should be placed within an area of

approximately 25x20 cm in order for all the cups to attach to the bag. To

ensure a good grip and so that the knife would not harm the construction,

the suction cups were placed so they would operate on the lower part of the

bag.

(28)

In order to ensure a functioning robot tool and a high safety factor, the number of suction cups were calculated with Equation 3.2 and the diameter of the suction cups with Equation 3.1. Results can be seen in Figure 5.7.

To further increase the safety factor, eight suction cups were used.

Calculations suggested a minimum diameter of 50,5 mm, closest available model was 62 mm.

Calculation of number of cups n Calculation of diameter D

F

th

= 54N D = 50, 5mm

F

s

= 20N m = 5, 5kg

n = 2, 7st c = 2, 7pcs

b = −0, 4bar s = 4

Figure 5.7: Results from calculations of diameter and number of cups neces- sary.

A vacuum switch, which triggers on the level of vacuum generated by an ejector, was mounted on one of the ejectors as seen in Figure 5.8. By trimming a potentiometer, the switch activates at a user-defined level. The switch was installed in order to determine, when the vacuum was on, whether the bag was in the tool or if the robot had mistakenly dropped the bag.

Figure 5.8: Vacuum switch functioning as a security check

(29)

5.4 Cutting station

Initial discussions took place regarding how the bag could be opened in a controlled and repeatably way. Two options discussed were to integrate a knife into the robot tool or to mount it on a separate rack within the robot cell. The choice eventually fell on the later option because of simpler implementation and increased personal safety when the blade movement is limited to a smaller area of the robot cell. Discussions also resulted in that the bag should only be cut in the woven side and not in the transparent, in order to maintain control of the bag after the cut had been made.

To ensure the repeatability of the cutting sequence, the knife was mounted on a pneumatically controlled linear unit which was to move alongside the bag, while the robot held the bag fixated in position. This way the bag could easily be stretched and smoothed out where the cut was to be made.

During tests of the setup, the linear unit was controlled by a manual switch. A construction, for mounting the knife firmly onto the linear unit, was drawn in SolidWorks and thereafter 3D printed. The printed

construction was done so that the cutting angle and depth of the knife were adjustable for later trimming. To ease maintenance, the knife was pinched by the printed construction and fixed by two magnets.

High frictional rollers were mounted directly above and below the knife, see Figure 5.9. These helped to make a clean cut. When the bag was pressed against the rollers they flattened out the area of the bag which was to be cut. Without the rollers the bag tended to fold slightly, complicating the cut. The linear unit was mounted in an offset angle from the rollers. The angle was experimentally configured and measured to a few degrees, as seen in Figure 5.10. By introducing an angle, a larger section of the blade could be used for the cut.

Figure 5.9: Linear unit and rollers

(30)

Figure 5.10: Linear unit, seen from above, mounted in an angle from the rollers.

Two different knifes were tested, one with a long straight cutting surface and one with a bent, hook-like, cutting surface. The test aimed to evaluate whether a tomato cut performed with a bent blade, see section 3.3, or a regular slicing cut with a straight knife would serve the application best.

The two different models of knives tested can be seen in Appendix A.

Results of the tests with different knives showed that, while a tomato cut partially worked, a purely slicing cut worked more frequently. In order to make a clean cut, i.e. not to leave any uncut material and without tearing the bag, the bent blade required a greater velocity than the straight knife.To only cut one side of the bag and not puncture the welded seam, proved to be harder with the bent blade.

Problems encountered during testing of the cutting station was that with poor adjustment of the supporting rollers and the printed construction, the knife would cut too deep into the bag. Results from a cut made too deep included that the knife slit the seam on the long axis of the bag, or that the knife also punctured the transparent side of the bag.

5.5 Robot tool assembly

To be able to mount and adjust the suction cups and ejectors, an aluminum bracket with cut out slides was ordered. Along with this adapters between the hole pattern on the ejector and the slides of the bracket was 3D printed, seen in Appendix B. The design of these components was a result of yet another iterative process with ideating phases. To get a fully functioning tool the bracket was bolted to the pinching tool.

Tests showed that the vacuum flow in the suction cups that were coupled

(31)

last in line decreased severely compared to those who were first in line, if the suction cups were coupled in series. To aid this the suction cups were coupled in parallel via an air distributor bolted onto the bracket, see Figure 5.11.

Figure 5.11: Air distribution of the ejectors, each pair is coupled in parallel.

5.6 Robot simulation

Using SRS, a robot cell was set up and a cycle routine was programmed.

The simulation is built upon a main task which instructs the robot in what movement to execute. A parallel task records changes in instructional variables, which triggers which action of the main task that is to be performed. To better reflect the thought robot cell, estimated delays were implemented in the code. A wait time of two seconds was inserted when the bag was to be cut and when the robot disposed the bag.

The production cycle is divided into four steps; picking, cutting, pouring and disposal of the bag. The picking action starts with the bag in a known fixed position on a small table. The robot moves towards the bag and activates the clamp. This enables the robot to pick up the bag slightly.

Vacuum ejectors are then activated to fixate the bag in the tool. The bag is

then transferred to the cutting station and gently pressed against the

structure. The linear unit holding the knife is activated. After the cut has

(32)

been made, the bag is moved to the pouring station. A tilting move is performed by the robot which causes the caps to pour out of the bag into a tray. The bag is thereafter positioned over another tray which serves as a recycle bin. The bag is dropped by first turning off the vacuum ejectors and then re-open the clamp. The robot moves to its starting position and the cycle is then complete, whereupon the robot is ready to pick up a new bag.

An overview of the robot cell constructed in SRS with parts from SolidWorks can be seen in Figure 5.12.

Figure 5.12: Overview of simulated robot cell.

(33)

5.7 Economy

In this project, the expected positive change in annual cash flow is

generated by reducing the cost of the labour. The whole project is expected to reduce one person per shift and the remaining staff is expected to be able to maintain the new automation as well as the current, if implemented.

The chosen robot is recommended by ABB Customer Service [18] based on estimated load capacity, reach and environment.

Cost of labour today: 500.000 sek/year and person * 5 shifts = 2,5 Msek/year and position [4].

Cost of investment: (This robot cell)

Robot ABB IRB 4600-45/2,05 475.000 sek [18]

Robot tools and other mechanichs 150.000 sek

Further development 1000 sek/hour * 500h

Implementation 1000 sek/hour * 1500h

Safety and surveillance 75 000 sek [15]

Total 2,7 Msek

Cost of investment: (Whole project)

This robot cell 2,7 Msek

Automated material flow 2,0 Msek [28]

Total 4,7 Msek

(34)

Chapter 6 Result

6.1 Final designs

The entire gripping tool was a combination of a clamp and a set of suction cups. Both parts were operated manually by switches and driven by

compressed air. The tool can be seen in Figure 6.1. The tool is able to grip and hold a bag with caps even when the robot is in motion.

Figure 6.1: Complete gripping tool.

Cutting the bag is performed in the cutting station, seen in Figure 6.2.

This part is not included in the tool and only cuts through one side of the

bag. The blade runs along a linear unit, is operated through a manual

(35)

switch and driven by compressed air.

(a) Complete cutting station (b) Zoom in on knife holder Figure 6.2: Complete cutting station (a) with zoom in on the knife holder (b).

How the robot test arena is set up can be seen in the panoramic picture in Figure 6.3. The picture is taken from where the robot is based and the robot is working clockwise. The bag is placed in a specially made slot so that it is placed in the same position every time, see Appendix A. The robot controller can be seen laying on the bench to the right of the bag.

The red bin is where the bag is being emptied and the green bin is for disposing the bag. Every task in the robot test arena is performed well enough to confirm a proof of concept. Pictures of when the robot is performing these tasks can be seen in Appendix A.

Figure 6.3: Panoramic view of the robot test arena.

(36)

The method used to grab and open the bag has been designed to meet the requirements of a C-classed clean room. Some materials used to fabricate components, like the carbon steel blade or the plastic blade holder, are not approved to use in clean rooms. However, these can easily be replaced with other materials if the robot cell is to be implemented in the factory.

6.2 Economical scenarios

Investment case 1:

Solely this robot cell won’t contribute to any positive cash flow, hence the payback time is eternal.

Investment case 2:

The whole project, expected to reduce one employee per shift. Using Equation 3.3 the expected payback time can be calculated to

4, 7M sek/2, 5M sek = 1, 88years.

This means the investment pays off in just under two years.

6.3 Robot simulation

The robot simulation showed that with a working speed of the robot set to 50 %, the cycle time for handling one bag was approximately 16 seconds.

This included the four steps; picking, cutting, pouring and disposal of the

bag, along with added time delays for operating the robot tool and the

knife. Today’s production uses approximately 15 bags of caps during an

hour of full scale production. This translates to a minimum active working

time of the robot cell to 16 s/bag ∗ 15 bags = 4 minutes per hour. An

overview of the simulated robot cell can be seen in Figure 6.3 in section 5.6.

(37)

Chapter 7 Discussion

7.1 Is automation the way to go?

Our thesis is a part of a larger project aiming to reduce the workforce at Fresenius Kabi’s factory in Uppsala, replacing part of the staff with robots.

As it looks like today; society, the market and the economy has a strive for ever going growth. A direct consequence of this is that large corporations and companies are constantly trying to reduce costs and enhance

productivity. This might seem wrong at first glance to many people,

especially to those being laid off and the people closest to them. But if the alternative is to move production to countries where labour is cheap but the working conditions are dreadful and the standard of living is low? Then automation at the factory might not seem too bad after all. On the other hand, moving industries to third world countries might be a good way to improve the living standards in these countries if you look at the long-term consequences.

No matter your opinion in this matter, there is no denying that technology and robots are an increasing factor in modern industries. As long as

nothing very radical happens in the world, there is nothing stopping this

development so we might as well embrace it. If we learn how to use the

advantages of automation and technology it might be the one of the

answers on how to keep the welfare in countries like Sweden, increase the

welfare in other countries and still reduce the urgent threat from the green

house effect. Because positive outcomes of higher welfare is that it gets

easier for people to get a good education and to make choices that are

beneficial for the environment. It also tends to lead to a reduction of how

many children the average family gets. This will have positive effects on the

(38)

excessive human population in the world that we can see today.

To resolve the matter of an aging population in some countries, automation is a necessity. Partly because there are fewer inhabitants within working age but also to be able to increase the working age. This can be achieved by making robots perform the tasks that are monotonic, heavy or in other ways tough for the human body to perform repeatedly.

7.2 Vacuum switch

In order to raise safety and control over the automated production a

vacuum switch was installed on the robot tool. The switch was mounted on one of the center ejectors. Only mounting a switch to one ejector seemed sufficient since this will likely be the last ejector to lose vacuum if the bag would be dropped. A larger set of switches would increase the cost of the robot tool without noticeably increasing the safety control. Perhaps could another switch act as a back up security check, should the first switch be malfunctioning. The switch was however never tested during the project but thought as preparation for future implementation. Once the robot can work automatically, the switch will function as an observer to make sure that the production cycle works as intended. The switch would alarm the control system by switching a digital signal in case the bag would be accidentally dropped.

7.3 Gripping the bag with vacuum

While results showed that a pre-fabricated suction cup was the best solution for gripping the bag, we think that a set of suction cups of model similar to 5.5c in Figure 5.5 might work as well. The suction force created by this model was convincing, although the design showed to be sensitive to jerking motions. With an alternative design of the bracket on which the ejectors were mounted, the plastic foam model might have proven to be sufficient. The main reason to why we did not follow through with this model was to maintain a confirmed and high safety factor. During

production in the factory it is crucial that the tool is functional at all times,

as an eventual production stop could be very costly.

(39)

7.4 Choice of suction cups

When evaluating a design, an important factor is the safety factor. An insufficient safety factor could result in a failure during production which is costly. In worst case scenarios, it could even lead to severe injury of

personnel. It is therefore crucial that the safety factor is not set too low.

When choosing suction cups for the robot tool, calculations resulted in that in order to maintain at least a safety factor of 4, the cups should measure 50,5 mm or more in diameter. A slightly wider model of cup was chosen to increase the safety factor. The same calculations showed that at least 3 cups should be used. In reality, one or more suction cup/vacuum ejector could fail due to unexpected circumstances during production. Since the calculations did not take this into consideration, 8 instead of 3 suction cups were mounted. This action generates an even higher safety factor. Tests conducted on the suction tool did not include any measurements regarding eventual centripetal force nor angular acceleration acting on the working object. The generous safety factor is imagined to compensate for this.

Another aspect of the increased safety factor is that it brings a larger flexibility when it comes to robot programming. When planning the program, the programmer does not have to be as cautious as if the safety factor would have been set lower.

7.5 Equipping the robot cell with a PLC

Since this project will only use one robot, the decision was taken to not implement a PLC into the robot cell. Instead, all communicative signals would to be handled by the robots control system. Should an alternative design instead show two or more collaborative robots that were to integrate with each other, a PLC might be suitable. While the Stäubli robot

controllers comes with the possibility of syncing between several systems,

not using a PLC as a hub would probably difficult the programming. It is

convenient for the programmer to think of the whole system as one master

unit with attached slave units. The introduction of several robots would

most certainly bring a larger set of signals for a control system to handle,

which is another reason for choosing a PLC, because of its larger capacity

and being more modular.

(40)

7.6 Manufacturing parts

A big problem encountered when manufacturing parts for the gripping tool was that the 3D printed plastic parts were too fragile for the application.

From the movement of the clamps, a considerable shock propagated in the tool and caused tear on the produced parts. This eventually led to several of the parts breaking. Keeping the design of the broken parts, models produced in aluminum proved to be more durable. However, where the forces on the parts were lower the printed models were kept. For instance, the final assemblage of parts forming the knife holder consisted of 3D printed material. This unit was considered too complex for aluminum milling.

7.7 Cutting station

Presently, workers in the factory use a pair of scissors for opening the bag.

This is considered to be a compliant method to use within a clean room.

However, integrating scissors in an automated solution would be difficult. It would be hard to mimic the motion of a human hand with a robot and maintain the precision needed. Instead of complicating things just to mimic today’s method, we aimed to find a cheap, easy and well functioning

method to open the bag. Especially since the whole robot cell needs to be tested to see whether or not it is acceptable to use in clean rooms anyway.

This was done in consensus with consulted experts.

A problem encountered while assembling the cutting station was that the cutting velocity was initially to high. With a too high speed the knife tended to rip the bag open instead of performing a clean cut. A solution to this problem was to adjust the in-built valves in the linear unit, which resulted in slower and smoother movements.

The selected blade used in the cutting station does not in its current form comply with the regulations of a clean room, as it’s made of carbon steel.

In order to make it compliant, a model of stainless steel needs to be found.

The knife presented in the report serves as a proof of concept but should

not be considered suitable for implementing in the factory.

(41)

7.8 Pinching tool

When designing the pinching tool, it was desired to place the piston as close to the moving parts of the clamp as possible. This was because we wanted to reduce the forces caused by leverage on the components.

Although we wanted as short distance as possible, it was also desirable to avoid unnecessary joints. The solution was to insert the piston in the fixed part of the clamp and mill a guiding trail for the connection rod.

A complication that occured with the clamp was that the clamping force was too weak to hold the bag in place . The main reason was that the piston we chose was too weak. Since several parts was already

manufactured to fit this piston, it was easier to enhance the friction between the clamp and the bag instead of change the whole piston to a more powerful. A different design with the active working force aligned with the clamps might have proven more sufficient and require less adjustment of clamping friction.

7.9 Repeatability

By performing the tests a few times we’ve proven that the concept works.

If this solution is to be implemented in a factory, the task needs to be performed thousands of times without any major errors. This demands that the construction is rigid and the programming is thought through and tested.

Our beliefs is that for this to be possible, the piston that operates the clamp needs to be more powerful. This would ensure that the bag would not slip out of the grip. But if the force of the clamp is enhanced, the empty bag might get stuck on the spikes of the clamp when it is supposed to be dropped.

The durability of what ever chosen blade also needs to be tested to

determine how many cycles it is able to withstand before it needs to be

replaced.

(42)

7.10 Flexibility of the robot cell

Using a six-axis robot was central in this project because it was provided at start. However, the question whether such a robot is the best choice for the application, or if a stiffer automation would be better suited, was brought up. By maintaining a higher level of flexibility the same robot cell may be used for several applications. Regarding this project, a stiffer automation might have proven as good or even better. However, the flexibility that follows a six-axis robot makes the robot cell less vulnerable for changes in production. A solution with high flexibility can, in general, be used for a longer time period with only minor modifications. Within the same period, a stiff automation might have to be remodeled from scratch in order to adapt to the same changes in production. However, since this project is conducted during a limited period of time, we chose to plan the project from the six-axis robot provided at start. While a stiff automation possibly would be the better choice, investigating this would be too time consuming and not fit for a project of this size.

The presented solution in this report is considered to have a relative high flexibility. It is insensitive for changes in production capacity because of the relative short cycle time along with few iterations required for opening enough bags. This leaves room for a greater production rate and enables the robot to perform various other tasks. Further, with the use of a tool changer the equipment flexibility may be greatly improved. This would enable the robot to be able to handle different models of bags or even cartridges.

7.11 Regulations of a clean room

A fundamental thought throughout this project have been to design a robot cell that in large extent could be implemented in a C-classed clean room at the factory. This have been decisive for choosing material and method of opening the bag. A large portion of designed components are either made of plastic or aluminum, both which are compliant with current GMP regulations. Due to these regulations, the bag was decided to be cut open and not ripped or torn open. A tearing of the bag could violate the GMP rules since it may release too many particles into the air. Performing a clean cut of the bag is therefore considered to be of great importance.

However, the robot model used in this project does not follow the GMP and

(43)

needs to be replace before future implementation in the factory is possible.

7.12 Progress towards an automated material flow

Should the larger project, introduced in subsection 2.2.1, be found worth implementing, an extensive amount of work would be required in order to equip the robot cell with a control system for surveillance of the cartridges.

This control system should be able to detect where full cartridges are stored and which ones need to be re-filled. One possible way of doing this is by using ultrasonic sensors. They could send an analog signal to a system which would calculate the distance from the sensor to the bottom of the cartridge. A full cartridge would thus generate a different signal than an empty one would. These signals could be used by the robot to determine which cartridges to fill, and by the AGVs to decide which ones to remove from the buffer. Should the distance itself prove to be unimportant for the robot cell and only determining whether the cartridge is full or not is sufficient, a sensor that sends a digital signal might be satisfying. This could be solved with an optical sensor along with a reflector, which would be triggered if the material level inside the cartridge exceeds a predefined level.

Another task that is to be solved before the robot cell can go live at the

factory is that the cell needs to be able to scan and keep track of batch

numbers. These batch numbers are currently marked on a sticker on the

bag and could be read with a bar code scanner. A system for finding the

bar code on the bag and transfer the information to a database needs to be

designed. A method of finding the bar code might include a machine vision

system that searches for the edges of the sticker. A less costly solution

would be to ensure that the stickers are always placed at the same relative

position and with a known orientation. That way, the system always know

where the sticker is once the robot has picked up the bag. Furthermore,

another bar code or a QR code could be implemented onto the cartridges in

order to track which cartridge containing which batch of material.

References

Related documents

46 Konkreta exempel skulle kunna vara främjandeinsatser för affärsänglar/affärsängelnätverk, skapa arenor där aktörer från utbuds- och efterfrågesidan kan mötas eller

Exakt hur dessa verksamheter har uppstått studeras inte i detalj, men nyetableringar kan exempelvis vara ett resultat av avknoppningar från större företag inklusive

The increasing availability of data and attention to services has increased the understanding of the contribution of services to innovation and productivity in

Av tabellen framgår att det behövs utförlig information om de projekt som genomförs vid instituten. Då Tillväxtanalys ska föreslå en metod som kan visa hur institutens verksamhet

Närmare 90 procent av de statliga medlen (intäkter och utgifter) för näringslivets klimatomställning går till generella styrmedel, det vill säga styrmedel som påverkar

Den förbättrade tillgängligheten berör framför allt boende i områden med en mycket hög eller hög tillgänglighet till tätorter, men även antalet personer med längre än

På många små orter i gles- och landsbygder, där varken några nya apotek eller försälj- ningsställen för receptfria läkemedel har tillkommit, är nätet av

The EU exports of waste abroad have negative environmental and public health consequences in the countries of destination, while resources for the circular economy.. domestically