Subproject D1: Modeling of Regeneration Supply Chains
This is a modal window.
The media could not be loaded, either because the server or network failed or because the format is not supported.
Formal Metadata
| Title |
| |
| Title of Series | ||
| Number of Parts | 6 | |
| Author | ||
| License | CC Attribution - NonCommercial - ShareAlike 3.0 Germany: You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal and non-commercial purpose as long as the work is attributed to the author in the manner specified by the author or licensor and the work or content is shared also in adapted form only under the conditions of this | |
| Identifiers | 10.5446/68357 (DOI) | |
| Publisher | ||
| Release Date | ||
| Language |
Content Metadata
| Subject Area | ||
| Genre | ||
| Abstract |
|
5
00:00
Capital shipModel buildingKette <Zugmittel>Playground slideRail transport operationsHammerFinger protocolSingle-cylinder engineRoll formingFinger protocolMachineRail transport operationsCapital shipPattern (sewing)Engine-generatorModel buildingTypesettingCartridge (firearms)Aircraft engineRailroad carSpare partComputer animation
06:26
Playground slideSingle-cylinder enginePhotographic processingModel buildingSuitcaseRail profileFinger protocolMovement (clockwork)Mint-made errorsTypesettingFinishing (textiles)Model buildingWorkshopAssembly lineMechanismus <Maschinendynamik>Spare partRutschungVolumetric flow rateShip of the lineToolComputer animation
12:53
Model buildingPlayground slideKette <Zugmittel>Mechanismus <Maschinendynamik>Reference workAssembly lineCasting defectKnifeAutomobile platformCapital shipSpare partStagecoachWatchStock (firearms)Ford FocusMaterialStock (firearms)Model buildingRutschungRemotely operated underwater vehicleSpare partMultistage rocketFullingAssembly linePattern (sewing)Roll formingMachineRail transport operationsReference workStagecoachKey (engineering)ToolShip of the lineMaterialFinger protocolFood storageTypesettingComputer animation
Transcript: English(auto-generated)
00:02
This video presents the subproject D1, modeling of regeneration supply chains within the Collaborative Research Center 871, regeneration of complex capital goods. The subproject D1 is focusing on production or regeneration logistics and by this forms an element
00:24
of the project area D that focuses on the economic related questions regarding the topic of regeneration. While most of the subprojects within the CRC are focused on aircraft engines as the typical product to be regenerated,
00:42
the regeneration logistics doesn't really need to separate or specify on a specific product, which is why the project D1 at first focused on assessing typical supply chain structures for a variety of products, starting from aircraft as well as their components like aircraft engines, rail vehicles and their transformers,
01:08
but also construction machinery or stationary products like wind turbines. All of these have in common that they are set together out of thousands of parts that are subject to deterioration during service or operation,
01:25
which leads to a limited lifespan or lifetime of each of these products that requires regeneration at the end of this lifecycle to bring the products back into service afterwards.
01:41
So, it is obvious that operators want to shorten the regeneration phase as much as possible, which corresponds to short turnaround in lead times and at the same time expect the highest possible punctuality by the regeneration service providers,
02:01
since they need to plan the regeneration process a long time in advance. These expectations in combination with the already mentioned high amount of uncertainty regarding the workload or the damage patterns to be expected from a regeneration product
02:23
leads to the demand for sufficient countermeasures to handle these uncertainties and overall the generation of robustness of the regeneration logistics process.
02:40
This leads to the overall goal of the sub-project D1, which is the improvement of logistics performance in the regeneration of complex capital goods by, on the one hand, reducing and handling of information uncertainty as well as modeling and the optimization of regeneration chains or supply chains and the processes within these.
03:07
So, these goals lead to a structure that also represents the structure of this presentation and that corresponds or includes three fields of action.
03:23
The first one is improving the forecast quality by means of Bayesian networks and this comprises data-based forecasting of future capacity demands of these regeneration orders. The second field of action is the assessment of planning quality and especially the evaluation
03:45
of consequences of these disturbances mentioned before on the planning processes and the planning quality. The third field of action in the end is the supply chain and design and configuration with the goal of the design of holistic and especially robust regeneration
04:06
logistics systems that are able to handle the high amount of information uncertainty that still prevails after improving forecast quality by means of Bayesian networks or other data-based forecasting procedures.
04:23
So, starting with the first field of action, the improvement of forecast quality by means of Bayesian networks, we developed a method or methodology that allows to predict future capacity demands of future regeneration orders
04:44
based on historic data from past regenerations as well as operational data from service, for example, like condition monitoring systems that continuously record sensor data during operation.
05:03
So, what we did was to structure these two streams of data within one overall, let's call it damage library, which forms the basis for the next step, which is the forecasting of future capacity demands based on the information stored in this damage library.
05:29
For this, we used Bayesian networks that are based on statistic relationships between input variables like the customer type or the region of operation and the outcome or output variable, which in our case was the expected capacity demand per regeneration order.
05:51
This forecast allows for a better or more precise capacity planning, which was the next step in the first field of action.
06:03
And we implemented this using mathematical modeling and split it up into a short, a medium and a long-term capacity planning module that allowed for a more precise capacity planning based on the forecast data we extracted using the Bayesian networks as well as the damage library.
06:33
As the databases for this, this was not only applied to aircraft engines, but also to the regeneration
06:42
of transformers for rail vehicles, as we saw as a typical product for regeneration on the second slide. And by this, we were able to add additional influencing factors to our damage library, like factors from more sophisticated physical
07:02
inspection or initial inspection, as well to prove the applicability of our approach to different or various products or product groups. The second field of action focuses on the description and assessment of
07:24
planning quality as the subsequent topic after the forecasting process we just discussed. And to allow for assessing and describing the planning quality, we developed the plan history diagram that is shown on this slide, which is based on the overall idea of a milestone trend analysis.
07:46
The plan history diagram describes the distance or the time distance between a planned date, for example, the planned day for shipment and the actual day of observation, which corresponds to a graph that is continuously
08:06
falling down towards the horizontal axis, as long as the planned date sticks to the same date over time. Once we see a planning iteration or a change in the planned date, we can observe spikes within
08:24
the diagram where an upward spike corresponds to a shift of the date towards the future and a spike pointing down side corresponding to the movement of a planned date towards or closer towards the observation date.
08:43
This means that once the graph touches the horizontal axis, the planned date is reached and the corresponding process should be finished. Besides these planned date shifts that we already discussed, you can also measure planning errors on specific milestones like shown
09:02
over here, as well as delivery delays if we know when or which date was communicated to the customer first. We already applied the plan history diagram to a regeneration service provider, where we compared four products of the same
09:23
type and normalized the plan history of their regeneration process to make them comparable, like shown in the diagram over here. As you can see, we found a very characteristic plan history within or around the milestone number 2, which corresponded to the finish
09:44
of the final inspection of the products and the corresponding planned date shifts due to the new information that was gained during this inspection. Overall, the plan history diagram allows for making planned date shifts visible and by
10:04
this one of the main causes of intransparency in regeneration planning processes and by this allows for the identification of potential for further improvements in these planning processes.
10:23
The third field of action focuses on the assessment and configuration of regeneration supply chains. To do so, we implemented an extensive simulation model in Siemens planned simulation following the generic regeneration process that comprises the initial disassembly of the products, their inspection
10:45
and the repair of damaged components, as well as the final reassembly and the testing process. To give you an idea of how the simulation model was structured below this macroscopic level, I would like to give you a view on the disassembly process we implemented.
11:06
Here we implemented not only one production principle, but two. The first one is the flowline principles that is widely applied in many MRO shops. And besides this, we implemented the construction site principle that is more and more used for large scale products that
11:26
do not fit on typical or conventional flow lines and need to be disassembled in a dock, which is called mostly. Another advantage of the construction site principle is that it allows for a very specific choice of the disassembly
11:47
depth and by this also the much more flexible reaction to the high amount of information uncertainty in regeneration. To make the simulation results easily available or accessible for regeneration service providers, we transferred the
12:08
simulation results to an easy-to-use tool implemented in the widely applied software Microsoft Excel. Overall, the simulation model allows for the evaluation of various configuration options for regeneration supply chains, with
12:26
the unique feature that it takes into account various regeneration specific disturbances, as well as typical reaction mechanisms like spare parts pooling that I would like to take as an example on the next slide,
12:44
where I like to show you the overall structure of the assessment tool we built based on the simulation results. This sums up a variety of key performance indicators for supply chain configuration assessment
13:01
on the supply chain level, like the average overall lateness of the regeneration supply chain, the average throughput time for the regeneration orders that are processed within the supply chain, or the overall capacity utilization, as well as indicators that are focused on critical processes within the supply chain, like lateness indicators regarding the repair and reassembly processes.
13:26
The diagram shows three major configurations. The basic configuration that is shown in blue forms the basis for the assessment of any other configuration based on relative comparison. The green configuration works as a reference to show that the tool performs in line with recent findings of other researchers on this topic.
13:48
The main difference between these two configurations is the choice of the sequencing rule applied in the repair process. While the blue one always prioritizes the order with the earliest due date, the
14:01
green reference configuration works on the basis of the first-in-first-out principle. As you can see, the green configuration leads to an increase in the average lateness as well as the scattering of the lateness on supply chain level as well as process level. This states that the tool works in line with the actual or most up-to-date research you can find in the scientific literature.
14:29
The third configuration is based on the basic configuration 1 but is subject to information uncertainty regarding the damage pattern of the regeneration orders as well.
14:44
The third configuration is based on the basic configuration 1 but is subject to information uncertainty regarding the damage pattern of the regeneration orders as a countermeasure single-stage spare parts pooling is applied to this configuration as well. While the uncertainty leads to a greater standard deviation of the output lateness in reassembly, significant improvements
15:08
regarding the average lateness as well as the share of orders delivered on time can be achieved. At the same time, the average throughput time stays constant and this affirms that these effects that have
15:23
just been discussed do not result from a structural change and thus making the simulation less comparable but from the flexibility gained using the spare parts pooling that allows to compensate for the uncertainties in regeneration.
15:41
Spare parts pooling forms a characteristic configuration option in regeneration supply chains which is why it is given special attention within the subproject D1 and especially on the next slide.
16:02
While the generic regeneration process represents a linear and order-specific process, regeneration supply chains were found to have a maximum of two pool stages that can be added to the supply chain. These allow for the provision of either serviceable or repairable components to compensate for disturbances caused
16:23
by information uncertainty or other influencing factors like product or machine breakdowns along the regeneration process. As already mentioned, the top logistics priority is the punctuality of the regeneration process which
16:40
is mainly set in the reassembly process that represents the convergence point of different material flows. To describe these converging material flows, the supply diagram has been extended so that it now can process inbound material streams from various material flows including the repair stock and the pools with different lateness deviations each.
17:05
As the overall output measure, it describes the disturbed whip that is waiting for reassembly but cannot be processed due to missing material.
17:25
This allowed to describe the effects of spare parts pooling on the disturbed whip by a universal operating curve shown on the right hand side over here. This describes the disturbed whip based on the average pool stock level available in the serviceable pool.
17:43
By this, the diagram not only supports dimensioning the pool stock levels for potential pool components but also the selection or prioritization of spare parts to be applicable for pooling since you can easily measure the effects of adding components to a pool or cut them out.
18:04
While this model is focused on single-stage pooling, the simulation also confirmed two-stage pooling, which means including a pool of repairable spare parts, to offer even greater potential to logistics performance improvement.
18:22
Based on this finding, future research is focused on the integrated modeling of two-stage pools or two-staged pooling and thus to gain the full potential of spare parts pooling. This is shown on the following slide. While this model is focused on single-stage pooling, the simulation also confirms two-stage pooling, which
18:58
means including a pool of repairable spare parts, to offer even greater potential to logistics performance improvement.
19:05
Based on this finding, future research is focused on the integrated modeling of two-stage pooling and thus to gain the full potential of spare parts pooling. This is shown on the following slide.
19:21
To do so, the effects of both of the pools on the component's availability as a metric like the service level of typical stocks or storages are first modeled separately before integrating them in a 3D dimensional model.
19:41
The difference between the upper and lower diagram is just this repair offset, which is due to the repair that is required if you swap components that require repair with components from the repairable pool, since these still require repair instead of the serviceable components that
20:03
can be immediately distributed to the reassembly and thus are available instantly. To combine these into a 3D model, if you transfer the repairable curve on the left-hand side
20:21
and then add the serviceable curve on the right-hand side as well as squeezed versions of this curve, this results in a characteristic surface that allows for the identification of possible shares of pool stock levels that in combination define the final availability of components to the reassembly.
20:48
And that allows to take on strategic choices, whether to stock serviceable or repairable pool parts, depending on, for example, the actual prices of both of these component types or quality levels,
21:05
as well as, for example, used or new parts availability on the market, which are only a few of possible influencing factors that need to be considered in the further research.
21:22
We are looking forward to transfer these results with a well-known regeneration service provider in a future transfer project that is mentioned on the downside of this slide over here.
21:40
With this slide, I am coming to an end of my presentation. Thanks for watching the video and have fun on the other project presentations within the final symposium.