Professional Documents
Culture Documents
Student Example 2
Student Example 2
Figure 1 RCA mascot “Nipper” Figure 2 1920’s horn loaded speaker Figure 3
horn driver (for Fig2) ..................................................................................................................................................3
Figure 4 moving iron speaker Figure 5 duel coil speakers with audio transformer
attached ......................................................................................................................................................................3
Figure 6 modern permanent magnet speakers ..........................................................................................................3
Figure 7, Celestion Vintage 30, voice coil to cone assembly Figure 8, Celestion vintage 30, voice coil,
spider and cone fitting to frame. 4
Figure 10 assembled and wired speaker excluding dust cap task h ...........................................................................6
Figure 11 pole plate task a..........................................................................................................................................6
Figure 12 pole and magnet task b ..............................................................................................................................6
Figure 13 pole magnet and top plate task c ...............................................................................................................6
Figure 14 voice coil task d...........................................................................................................................................6
Figure 15 voice coil and spider task e .........................................................................................................................6
Figure 16 voice coil spider and cone task f .................................................................................................................6
Figure 17 cone coil and spider in frame task g ...........................................................................................................6
Figure 18 proposed facility layout ..............................................................................................................................7
Figure 19 production floor basic layout....................................................................... Error! Bookmark not defined.
Figure 20 central robot parts loader layout. ............................................................... Error! Bookmark not defined.
1) Introduction
The aim of this work is to explore virtual system modelling to evaluate and improve the manufacturing system of
a low cost component speaker.
Modern manufacturing systems are often complex and highly automated, where seemingly small changes can
have significant impacts on productivity. Manufacturing system simulation allows for experimentation and
validation of process designs and modifications without the huge cost involved of altering existing systems.
However, some tasks do not lend themselves to automation. Intricate work with high variability, often on
soft or delicate parts still requires human dexterity. Accurate assembly of compliant structures by adhesive
bonding as found in speakers is one area where complete automation is challenging.
Figure 1 RCA mascot “Nipper” Figure 2 1920’s horn loaded speaker Figure 3 horn driver (for Fig2)
(Barraud, 1898) (Soyland, 2019)
Figure 4 moving iron speaker Figure 5 duel coil speakers with audio transformer attached
(UNKNOWN, 1929) (Carlson, 2020)
Figure 7, Celestion Vintage 30, voice coil to cone assembly Figure 8, Celestion vintage 30, voice coil, spider and cone fitting to frame.
(Greeves, 2012) Note voice coil guide fixture for critical feature alignment
(Greeves, 2012)
While there is not a great deal of academic articles on the manufacture of speakers available, a few
manufacturers have provided virtual tours of their facilities online, or at least provided some publicity shots of
facilities. A few striking similarities in the higher-end speaker manufacturers’ production systems are apparent.
The manufacturing systems seem to be based around a flow lines with manual workers performing the bulk of
the tasks such as the voice coil fitment and diaphragm assembly to frame fitting shown in figures 7 and 8 above.
This is typically aided with production fixtures to increase speed and accuracy of assembly.
General requirements
The manufacturing system for the speakers is required to output 1440 units per day, giving a line cycle time of
20 seconds, while sufficiently sized to fit into a building appropriate to the turnover of the business.
A lean manufacturing approach is suited to the low unit cost, to aid this the line needs to minimise work in
progress, maintain line balance and efficiency, and account for downtime based on equipment reliability.
Building size
Assuming that £2.00 of the £5.00 from the unit price is used for purchase of the pre-existing components, the
gross income of the company is £21600 PCW.
Based on average pricing of industrial property rent in Chichester (statistia, n.d.) and allowing a generous 15% of
turn over for rent, the factory footprint could be in the region of 590m2 (24x24.5m).
To meet the production targets in the longer term, an adjusted line cycle time is required to account for the
downtime of equipment and will be based on the time lost on an average year.
Assuming the factory operates for 50 weeks a year 8 hrs a day 5 days a week, leaving 2 weeks for planned
maintenance. (2000 working hrs).
This leaves 1920 hrs remaining in the year, (6914887 seconds) therefore the maximum line cycle time to meet
production target (360’000 units) over the course of a year is 19.21 seconds.
The line cycle time is equal to the longest station or bottleneck time.
Maximum line cycle time ≥ (n x station time)/n where n is the number of stations performing a task(s).
So any assembly task longer than 19.21 seconds must have more than one station or operator.
No tasks are sufficiently short to allow them to be combined at a station.
For this exercise, the spider is clamped between the frame and the top plate after the soft parts are assembled into the frame. The complete magnet assembly is then
attached to the frame allowing final positioning of the voice coil and pole piece to be done at this time. The dust cap is fitted later to allow access through the
diaphragm assembly to access the pole piece aligning feature. This is an approach like the Celestion speakers shown in figures 7-8.
Figure 13 voice coil Figure 14 voice coil Figure 16 voice coil Figure 15 cone coil
task d and spider task e spider and cone task f and spider in frame
task g
The proposed assembly sequence for the speaker is shown above (in Table 2), operation b, d, e, and j, all have a task time greater than the maximum line cycle time so
require parallel stations. The number of stations required = task time/ cycle time, rounded up to the nearest whole number.
3) Layout options
Using a proximity relationship table to guide the placements of the departments allows for a more efficient
initial layout process. The layout is modified until adjacency requirements are met.
Table 3 Proximity requirements of departments
Receiving E2 A A U A E4 U
Production E2 X E1 I1 E2 E2
Toolroom E3 U A X E4
Office X U U
Break and X U
hygiene
Shipping X
As can be seen in figure one, the departments that need to be adjacent are adjacent, with at least one
department between those that must be separated.
3.2 Production system layout
It is desired that the production line be automated as far as possible, however human operators are present, so
some layout considerations regarding access to equipment for repair and safety are required. Shipping and
receiving ideally need to be situated close to the beginning and end of the line, respectively. Beginning with the
robot parts loader centred line.
Both proposed production systems are arranged in a concertina fashion to reduce the length of the line and
reduce distance between stations. The two initial branches for assembly of soft and hard components joining
midway through.
To allow access to equipment for service and repair in the more complex robot loaded system, high-level
transport conveyors, above 2m, give foot passage underneath as well as allowing conveyors to cross at different
hights.
The production line is focused around the central parts feed robot, that supplies most components to the feeder
conveyors of the automated and manual assembly stations shown above in figure 19. This line occupies an area
of 62x24m including warehouses. Far too large for the turnover available and needs to be reduced to around
half. Conveyors of lengths that are a multiple of 5m is proving extremely difficult to implement in the space
available.
By contrast the production line equipped with manually loaded parts feeders is 46x23m and could be reduced
further. Significantly fewer feed conveyors are needed as parts are not being distributed from a central point.
Reduced complexity also gives some improvement in downtime predicted.
Figure 22 3D view of hand replenished line
Conveyors do not need to be run at significant distance of the ground as access to all stations is not impeded.
The green availability controls shown in figure 21 above are not a physical barrier, but rather an artefact of the
program interface, the 3D view above in figure 22 shows the access room available.
Figures 21 and 22 above shows the availability controls that simulate the reliability and downtime expected.
Rather than assign control to every conveyor, control is grouped such that in the event of failure the section that
needs to shutdown to safely repair on the damaged item. Figure 23 below shows one example of this where loss
of the loader robot results in shutdown of conveyors to and from it to effect a safe repair.
By adding an additional short buffer conveyor intermediate to the dust cap bonding station production could
continue for a short period.
Figure 26 conveyor zone and magnetizer reliability
Where a station or process is operated in parallel such as the magnetiser stations shown above in figure 22 their
downtime is assigned independently to accurately simulate the low probability of simultaneous failure.
Hand loaded
5 year
Robot loaded
1 day
Robot loaded
5 year
The simulations are also run over a 2 working week cycle to look at shorter term output.
The system is required to produce 1400 units per working day, giving a 5 year output of 1.8 million units.
The robot centred system produced a total output of 1771751.60 units 28’294 units short (1.56%) of the desired
amount despite appearing to have excess capacity, this equates to £84’747 in lost turnover.
The Human loaded system produced 1767188.80 units 32’812 units short (1.82%) of the 1.8 million units, a
turnover lost of £98’463 over the 5 year period.
This contrasts with the expected daily outputs when simulated over 10 working days where both systems on
average appear to over-produce.
Both the hand loaded hopper system and robot loaded variant of the production line have their individual merits
and could be fine-tuned to give the required production volume long-term, with small changes. The preferred
method really depends on the cost of floorspace available versus the wages of the workforce required.
There is some anecdotal evidence and personal experience to suggest that a “leave that for the day shift”
attitude amongst night-shift workers can cause a surprising amount of disruption, adequate supervision and care
needs to be taken during the night-shift to control this issue.
Further refinements to both simulations would be beneficial to improve the accuracy of the results.
At present the feeder robot and buffer conveyors place parts at random, due to issues with the system locking
up when identifying individual random components. Correcting this would allow for a better understanding of
the length of the buffer conveyors required to ensure adequate capacity to absorb the random variation.
Alternatively, an additional robot loader could be used with the part types distributed between them. Provided
that sufficient buffering is present the output should not be significantly affected.
Further optimisation of the layout of the system, conveyor length and capacity could reduce footprint and work
in progress considerably while increasing uptime on the production line.
The human loaded feed hopper simulations do not accurately reflect the component storage and refilling
methods. Storing 1500 components, even relatively small ones is a non-trivial task. It is likely that replenishment
while the system is online would be required. Magazine or tape feeding of components is commonly employed
in the electronics industry, along with automated pick and place. This includes systems for online replenishment
of components, while this would add complexity, this has been accounted for in the MTBF and MTTR of the
feeder systems in the model.
To reduce footprint, the hard parts line’s first three stations do not have a buffer conveyor between them, or
between the station and the parts feeder system. While this has little effect during normal operation, any
interruption of feedstock results in immediate loss of production. Adding buffering would allow production to
continue until the buffer stock is exhausted, giving an increase in productivity. The size of the buffer is a trade-
off between work in progress and productivity gained during feeder failure.
Addition of an accurate 3D model of all components and sub-assemblies to the simulation would allow a more
natural understanding of how the process is progressing and what is occurring at each station. This has not been
implemented due to a quirk of the remote access licences of the CAD and Enterprise Dynamics software used
not allowing import of DXF or export of .wrl files. This however does not impact the simulation results in a
meaningful way.
Neither system is compact enough to fit within the floorspace that the turnover of the business would provide,
this is partly due to the conveyors being of 5m increments in length. Alterations of length and capacity of
conveyor systems would facilitate reduction in footprint area of the systems, along with a reduction in work in
progress.
Raw data from one test given in the appendices shows considerable time spent by components in some buffer
conveyors, these could have their capacity reduced to bring down in process inventory.
Given the cost of commercial rent I would be inclined to use the hand-loaded system unless further work
indicates cost savings from the alternative.
References
Barraud, F. (1898). Retrieved from
https://en.wikipedia.org/wiki/Horn_loudspeaker#/media/File:OriginalNipper.jpg
Carlson, P. (2020, 04 30). Electronic Repair Adventure- Radio Receiver Troubleshooting. Retrieved from youtube:
https://www.youtube.com/watch?v=_11CkjXT2nw
Soyland, R. (2019). RESTORING AN OLD AMPLION HORN SPEAKER. Retrieved from youtube:
https://www.youtube.com/watch?v=azEFQXpgynA
UNKNOWN. (1929, june 29). VOLUME 15 NUMBER 15. Radio world magazine, p. 21.
Appendices
End of report.