Automation – Analyzing a Cobot CNC Machine Tending Application
As with all machine tending applications, the automation of the CNC machine load/unload functions has the dual objective of minimizing/eliminating labor while simultaneously getting the highest utilization of the machine in question, capacity.?In this context, keeping “the Spindle running and making chips” with the theoretical but impossible limit of “zero” load-unload time.?In general and assuming that it has been optimized during the CNC programming phase, the cycle time of actually machining the part cannot be changed and what we are left with is to minimize the finished part removal and introduction of the next part/stock.
We might say that the load/unload function, while necessary, is “non-value add” time while the machining function is “value add” time.
Most recently, this application has been often addressed with the use of Cobots, tooled with appropriate end-effectors, (grippers, vacuum cups, etc.), which standardize the total cycle and make it both faster as well as more reliable than a human operator, who is now able to attend to perhaps more than one system by loading parts/stock into a fixture or conveyor that presents those parts in batches to keep the Cobot replenished and running.
Having seen many videos of these applications, utilizing various Cobots, and with substantially the same sequence, it becomes clear that while the Cobot offers significant improvements on both the elimination of labor and spindle utilization criteria, there appears to be some significant “meat left on the bone” ?in term of total productivity improvement which are a combination of:
A) The robot sequence logic using “serial” wait times in the load/unload sequence, and
B) The inherent limitations of the Cobot having to go “slow” in order to conform to it’s “cobot” status in a particular application.
Analyzing a representative video of a typical application on YouTube, (taken a random and one of the better ones seen), we can make some observations to see, How? if Possible, we might increase spindle time ON while increasing the throughput per hour, by some economically meaningful amount. (See short analysis below.)
The YouTube Video is Here:
In this example we observe the following sequence, starting with the CNC door OPEN:
1.??????Cobot , (holding the stock/next part), going from the part/stock of the matrix fixture, reaching the vice, un-clamping, retrieving and blowing off the just finished part, placing and clamping the next part in the vice, coming out and clearing the door so it can close is 28 seconds per cycle.
2.??????The Door Open to Door Close to Door Open Cycle Time?is 30 seconds per cycle – (This is the Total CNC Time and represents the theoretical limit of how fast we can go)
For a Total Cycle time of 58 seconds
This yields roughly – 62 parts per hour with the CNC machining time at 52% available capacity.
Potential Improvements
1.??????We observe that when the Cobot has retrieved the next part/stock while the CNC is processing the current part, it is waiting over the parts fixture/matrix until the door is fully open.?(Presumably it receives a signal when that has happened or perhaps it’s simply based on a timer based on some previous signal.)?In any case, that takes 6 seconds, owing to the particular sequence and the speed limitations of the cobot, to reach the point where the door “plane” is located.?
If instead of staying parked over the parts fixture, the Cobot holding the next part moves to the door plane while the CNC is running and waits there, we can take that 6 seconds out of the total load/unload cycle and that time now becomes 22 seconds.?
This changes the output per hour from 62 to?69 pieces and this is a 12% increase from the original baseline with now 58% of available CNC machining capacity being used.
2.??????Having perhaps captured the low hanging fruit in (1), we further observe that once the Cobot reaches the vice, it takes about 8 seconds to unclamp and “blow off” the part, removing chips first from the part and then also the vice. ??This is clearly a vestige of how a machinist would manually perform this operation and this impacts the cycle time as it may be better buried in the machining cycle.
To recapture most if not all of these 8 seconds, it would be quite simple to add air nozzles to the spindle above and/or the vice itself, similar to the coolant “goosenecks”, whereby once the machining is complete, the air would turn ON, whether controlled by the CNC or Cobot controls would be a matter of preference and convenience and would happen in parallel with the spindle retracting UP and the Door opening at the end of the machining cycle. If we take 6 of those 8 seconds saved, our load/unload cycle now becomes 16 seconds.
This now increases the output per hour from 69 to?78 pieces and this is a 26% increase from the original baseline and now at 65% of available CNC machining capacity.
3.??????Finally, perhaps we can “finesse” a couple of additional seconds out of the Cobot sequence/movements which are relatively slow as well as being “rectilinear” so that by rounding some corners and minimizing some sequential wait delays, we can reach a load/unload cycle time of say 14 seconds. (Note: these are only the movements required to get the Cobot into and out of the CNC.)
This now changes the output per hour from 78 to?82 pieces and this is now a 32% increase from the baseline and now at 68% of total available CNC machining capacity. (If we can accomplish this, we will have in effect created 1/3rd additional CNC machine for this part which as a practical matter, we can, depending on quantity required, dramatically impact job scheduling flexibility, better meet customer delivery requirements and in general improve our productivity without increasing our manufacturing asset base.
For our discussion purposes:
-?????????CNC Machining Capacity is defined as:
((CNC Machine Cycle Time)/( CNC Machine Cycle Time) + (Unload/Load Time)) ??
-?????????CNC Machine Cycle Time = Total Time from Door Open to next Door Open
***??????100% CNC Machining Capacity would mean that the spindle never turns OFF and we are able to change-out a finished part with new stock in 0 seconds. (Not real of course)
For reference, if the machine load/unload cycle, in this example, can be made 4 seconds, that would increase our output from 62 to 106 pieces per hour, and this is a 71% increase from the baseline and now using 88% of total available CNC machining capacity for this job.?This is unlikely to be achievable with a Cobot but thinking along those lines of maximizing the spindle ON time, might spur ideas for a flexible parts handler for this common application which is cost justifiable as well as meeting the productivity objectives.?In essence, such automation would be creating almost a 2nd CNC machine!
领英推荐
The Cobot automation solutions have been adopted by hi-mix low-volume manufacturers and the advantages over manual load-unload can be significant.?In addition they may provide some reprieve to the scarcity of labor in some cases while allowing backward/forward compatibility to use the CNC in either auto or manual load/unload mode as may be required.
FINAL THOUGHTS
1.??????While the ROI is reported to be met in several months in some cases, (although the calculation methodology used is not clear), using this form factor example, it appears that implementations are not always as efficient as they might be in order to impact productivity to a greatest extent possible.?This may be due to the possibility that those deploying these solutions lack automation experience or that they miss the fact that the objective is to have the CNC “making chips” for as a high a duty cycle as possible.?In either case, this implementation is certainly a big step forward and more consistent than manual load-unload while moving the labor required one step further back from the CNC itself and “off-line”.
2.??????While there are many testimonials on Cobot manufacturers’ websites of successful implementations and success stories, these typically compare the “before”, manual operation to the “new”, Cobot load/unload operation and the overall impacts on total productivity are not yet fully documented.??Undoubtedly these are still early days and the manufacturers themselves are learning as they adopt this approach more widely.
3.??????In the example shown we did not touch the CNC cycle itself but the doors opening and closing are 5 seconds each or 10 seconds total and included in that total cycle.?This represents 1/3 of the total CNC cycle in this job.?Depending on the total machining time, this can be a significant portion of the total time, as in this case, and an opportunity to significantly reduce or mitigate this time would further increase productivity as we can see that every second of non-value add time significantly impacts the output.?(Keeping in mind the doors are a safety feature and that attribute cannot be compromised with any changes/improvements.)
Improvements to the door may be accomplished by:
-?????????A better motion coordination via sensors to start the Cobot moving while the door is still closing and/or opening.
-?????????Limiting the door Open-Close stroke to the minimum needed for Cobot access, thus lessening this time when the Cobot can start moving into the CNC.
-?????????Changing the door open-close drive to significantly reduce the open-close times, perhaps by 50+% with a robust “smart” drive that controls accel/decel.?(Initially seemingly costly but more than worthwhile in terms of increasing machine utilization and throughput as we can see.)
4.??????It also appears from this rough analysis that the value proposition is not in using the Cobot, per se, but in minimizing the time of retrieving the machined part and introducing the next part/stock into the vice while clearing the door: in and out QUICKLY! so the spindle can start spinning again.
5.??????The “place” position of the “pick and place” Cobot function is fixed, while the “pick” position is variable, but that’s only because the Cobot has the ability to pick from any of the locations on the part fixture/matrix.??Since the matrix is loaded by a human worker, it follows that he can just as easily be loading an indexable carrier that always presents the part to a fixed “pick” position.?Therefore, using the Cobot in this application may or may not be the most efficient way to achieve the goal of minimized time IN-OUT parts retrieval and presentation.
6.??????Having reviewed many versions of this embodiment of this application, they generally fall into two major groups.?Loading parts, (cylindrical or rectangular), of limited mass, into a vice or into a spindle located to either side of the CNC machine.?As such, the “place” location is always fixed while the “pick” location is usually variable from a tray or a matrix fixture.?Since the main objective is to reduce the load-unload as well as close the doors as soon as possible to get the spindle spinning, it’s clear that any/all “seconds” of delay significantly impact the total machine utilization time and should be minimized.??Minimizing these delays is the real value proposition!
While not seen in this example, some examples include steps like the Cobot Opening and Closing the door as well as hitting the “Start” button on the machine.?Those are all needless delays that can be avoided by automating those steps, again either through the Cobot or CNC Controls.?Deploying this type of cell is a significant investment in hardware and training and leaving those steps to the Cobot to mechanically perform, (mimicking the human worker in effect), is being penny wise and pound foolish! Given the productivity penalty paid.
7.??????The use of Cobots in the tight/crowded machining environments, so that workers can be in the area and can present new batches of parts has also served the singular and important function of introducing automation to this long neglected niche.??While the CNC programming is still required, the use of that same higher skilled worker to perform this relatively low skill job of “load” and “unload” has proven to be both cost prohibitive as well as difficult to staff over time.?This has now been recognized by most high-mix, low-volume CNC machining manufacturers, whether captive in-house shops, doing regular low quantities, (100 – 5000 pcs) or general contract shops dealing with similar, infrequent but regular orders of similar quantities.
CONCLUSIONs
As a first pass, the Cobots for this application have both proven the automation possibilities but have also demonstrated their limitations in terms to maximizing the spindle run time as they are basically automating the manual worker NOT the process.?Where the Cobot has shown greater benefits is when it is part of an integrated cell which may use the machining cycle time to perform other operations, such as placing parts in an ultrasonic cleaner or perform some secondary operation like presenting the parts to a laser marker or other final process steps.
This application appears to have a niche that may be substantial to productivity improvement, IF the “pick” location can be made constant via a manually loaded indexer/carrier and the actual delivery function reduced to a four axes, (X-Y-Z-θ), short stroke mechanism that takes the Next Part from just outside the CNC door into the Machine/Vice while retrieving the just completed part and returning it to the same pick location.?Aiding this effort to minimize strokes and part replacement, the CNC itself may be used to a greater benefit via moving the vice closest to the door and/or using the spindle itself in taking the just finished part out of the vice, further shortening the “IN-OUT” time.
Meeting the safety requirements, such a mechanism would be guarded and “stick” out into the CNC machine at whatever speeds required to maintain both accuracy and minimizing the unload/load time via shorter strokes.
From a strictly “Lean Manufacturing” perspective, most of the motions as performed by the Cobot in the representative example shown are non-value add and any illusion or “flexibility” shown is primarily due to the layout of the inputs … itself the result of the flexibility and/or limitations of the Cobot.
In either case, when automating the CNC unload/load function the prime design driver should be the minimization of the Door Open to next Door Open cycle time in order to maximize productivity which after all is the main reason for automating any manufacturing process.?
Focusing on ROI alone does little to innovate the manufacturing process or improve competitiveness.
NOTE:? The representative analysis presented here and the video selected are for illustration purposes, as these deployments have pretty much the same sequence steps, and as such is not meant to be either an endorsement or a critique of this system as the details and the background constraints that went into designing this example are not known to the writer. ?
Using this or a similar type of analysis, Cobot machine tending deployments can be reviewed to identify and capture any further productivity gains that may be “meat left on the bone”.
Your thoughts are appreciated and please share this article if you think your connections will find it of interest.
Connect or DM to discuss further.
#manufacturing #training #development #reskilling #upskilling #apprenticeships #workforcedevelopment #capitalequipment #scientificmanagement #factoryofthefuture #assembyfixtures #robotics #processautomation #automationsolutions #automation #strategicmanagement #roboticautomation #lean #manufacturing #cobots #assembly #packagingautomation #leanmanufacturing #industry4 #ai #rpa #artificialintelligence #innovation #assembly #industry40 #efficiency #digitaltransformation #industrialengineering