What Is It?

Supply chain simulation is the off-line alteration of master data and sometimes transaction data in order to see how changes effect the output of the plan (separated into planning models and versions in SAP SCM).

Separate Hardware or Same Hardware

Simulation can be performed on a separate “box” from the production server so that it does not interfere with the production environment. Or it can simply use the same hardware but have a second or third or fourth database. In a mixed environment there are more capabilities to run planning without worrying when production is running activities. However, when attempting to migrate changes from the simulation version, the official SAP position is that it can be easier if the same box is used for both production and simulation. However, this is not necessarily the case as the mass maintenance function can be used to take changes from a simulation box to the production box. As we will see, because simulation is so rarely implemented with APO (due to overhead and problems APO has in terms of flexibility), its difficult to come to any strong recommendations either way.

Why Simulation is So Important

Effective simulation is often the missing ingredient for successful advanced planning projects because without it, hypothesis can not be as easily or as robustly tested. Most companies have a number of viewpoints of planning that they have developed over time, however these viewpoints can be valid or invalid, but testing is an effective way to find out how valid they are. That is just the beginning, with a simulation capability (not just the software, but trained individuals, and a company which is open to simulation, etc…) ideas can be tested that the business may not have thought of.

Supply Chain Digest has the following to say about simulation:

That is where “simulation” can come into play. In simulation, a model of the system is built (again, whether it’s a conveyor system in a DC or a supply chain network). Rules are created (often still through programming, but increasingly with at least some level of system configuration) that describe how the system should work. – SCDigest

Simulation with Advanced Planning Systems vs. Simulation with Spreadsheets

“What if” analysis is traditionally performed with spreadsheets, at least since their development several decades ago. Often the question is raised regarding the how the advanced planning simulation is an improvement over spreadsheets analysis. There are many advantages provided by an advanced planning simulation system. The first of which is that a simulation environment can tell its users the implications of changes not only in isolation, but in terms of how they affect other parts of the supply chain.

_________________________________________

Implications Between Locations and Multi Echelon

This brings up an important point. Simulation can only simulate up to the functionality of the planning system. For instance, SNP in APO can simulate what happens how inventory level changes for a group of materials when the service level is changed (if SNP is using enhanced safety stock methods). It can do this for the entire database of materials. However, because it is not a multi-echlon system, it can not show how changes in stocking policies at one level in supply network affect other areas of the network. This is because SNP plans all locations sequentially. I am increasingly of the view that supply network planning software should be multi-echelon, so this makes me less enthusiastic about performing simulation in SNP. However, that is more of an exception. Simulation is still very useful in the other modules such as PP/DS, TP/VS, etc..

_________________________________________

Secondly, there are many items that a simulation can test, that can not realistically be modeled in a spreadsheet. When performed correctly, the simulation environment provides a more robust solution, with higher confidence, particularly because the simulation is running with most of the production model parameters in place. This point works in conjunction with the first part regarding the comprehensiveness of the analysis that simulation allows for.

Simulation with SCM

The transaction /SAPAPO/MVM is what is used to get into simulation. This is where planning models and versions can be copied, deleted and otherwise manipulated. According to SAP, some things can be tested in simulation, that can not ordinarily be performed in the active version.

Production Planning and Detailed Scheduling transparently handles multiplant manufacturing environments. You can add interplant transfer requirements and planned orders in simulation mode, with plans set up to run automatically or interactively. You can develop plans for interdependent plants independently with constraint violations established and propagated as alarm conditions, or you can simply build a comprehensive plan that views the processes of multiple plants as seamless members of the overall production process. – SAP White Paper – Production Planning and Detailed Scheduling

Is SCM A Natural Simulation Environment?

It should be understood that SAP SCM is not the best simulation environment. SAP made a number of decisions regarding the design of SCM that move it away from simulation. Good simulation environments are often built around a “black box” optimizer (SCM is often run not using any optimization), use an easy to alter relational backend (SAP requires specific transactions such as MASSD, which is more clumsy than a SQL front-end that most other advanced planning applications use). SAP SCM takes a lot of resources simply to run normally, which is not the desired state of a simulation environment. Finally migrating through the SAP SCM interface, compared to others that we have used is slow. We have listed positive simulation criteria below:

• How quickly can the model parameters be changed
• How good are the reports of the simulation runs
• How easily can data be migrated between the simulation and active version

However, while SCM can not be said to be a natural simulation environment, it is clearly beneficial to use the same software that one uses for the actual planning that is used for simulation. One reason is that there is normally the desire to migrate the changes from successful simulations back to the production version. SAP SCM has a complex copy method that can be used to do this, however, depending upon what is changed, again, the MASSD transaction may be a more straightforward may to “migrate” the changes from the simulation version to the active version.

How Simulation is Presented

Much is discussion on simulation. This is a frequently desired functionality. However, less often discussed is how simulation is rolled out.

Simulation has found wide acceptance in a number of fields. However, supply chain simulation is more discussed than actually performed. This has primarily to do with education and the resources that are required to perform simulation. Software is very rarely demonstrated during the sales phase in terms of simulation capability, so a vendor like SAP has less incentive to drive the product in that direction. The highest level of simulation is when there is the ability to document simulation runs and to effectively socialize the results in a meaningful way.

How Difficult Is Simulation?

Simulation is not at all easy, although as we have previously mentioned, software design has a strong effect on the ease of alternation of the supply chain model, as well as the ease of analysis of the simulation. However, even in very well designed simulation environments, previous exposure and experience in analysis and testing as well as documentation is required to effectively leverage the software selected. This is echoed in the white paper Supply Chain Simulation Modeling Made Easy: An Innovative Approach:

Simulation modeling and analysis requires skills and scientific background to be implemented. This is vital for this powerful methodology to deliver value to the company adopting it. There are several practices to implement and rely on simulation modeling for strategic and operational decision-making, including hiring simulation engineers, building internal simulation team, or contract consultants. – 2007 Winter Simulation Conference

We categorize this into the following areas:

• Computing resources
• Analytical resources, most often a dedicated analytical group
• The ability to document and archive the simulation runs
• The ability to socialize the results of the simulation

A Simulation Archival Blog

What is necessary, but as far as our research into this showed, very rarely discussed, is in addition to running the simulation, its important how the simulation results are documented and distributed or socialized. This way there is a record of the simulations that go back from the first simulation and continue up until the present time. Too often papers and conversations on simulation center around how technically accurate the simulation is, or the mathematics of simulation, without any focus on the application of the simulation results, or how the results are kept and made available. This relates to tools selected, but also to how a simulation group is staffed. Communication and information management as well as analytical capabilities need to be present in the simulation group in order to facilitate the migration of knowledge from the simulation group to the pertinent parts of the organization that are essentially customers for this analysis.

Setting up the changes to the model and performing the quantitative analysis is only one part of maintaining a robust simulation capability. The other part is documenting, categorizing and socializing the results.

The Simulation Blog

One idea I recently came up with is to house the results of simulations in a blog, where each simulation run is a different post.

______________________________________

Why Blogs Work

I have found blogs to be such effective information management tools that I have a wide number of them for different purposes. However, before blog software became available, I often struggled to keep a much smaller amount of information properly organized on a number of websites using software like Dreamweaver or other HTML editors. The unknown feature of blog platforms like WordPress, at least to non-bloggers, is that blogs are actually lightweight content management systems. The same software that manages sites like the Huffington Post or even some celebrity site, can be used to host any category of content from the most technical to travel photos. I now create a new blog for every account I work at, and it provides me with a searchable database of activity at the client. At one of my recent clients, they actively use the blog that I have setup for the project.

_______________________________________

Tags and Categories for Simulation Runs

This simulations can include screen shots of the system, and or attachments such as spreadsheets of extracts of SCM or offline analysis of the simulation result. These posts can be tagged in the blogging software such that they are highly categorized, so that a person searching for blog results can be easily searched. In fact, to get a feel for what I mean, notice the tags and categories to the right of this article. This blog you are reading is also served from WordPress software. It categorizes SAP SCM articles, however the same software can be used to categorize simulation runs. For instance, lead times could be a tag, so all posts which have that tag would appear when the tag is selected.

Secondly, in addition to tagging, blog software has strong search capabilities as well. This blog is hosted at a WordPress server, but if hosted on a private server, much more powerful search plug-ins can be installed which provide great searching capability. We would recommend very strongly against using SharePoint for this type of archival as it not designed for it, and would end up encapsulating rather than revealing the results and would be more difficult to administer, and could not as effectively socialize the results of the simulations.

What Is To Be Documented?

The simulation group should be setup with a number of areas of every simulation that must be documented. What is documented should remain consistent for each simulation run in order to maximize the comparability between runs.

For each run the following should be documented:

• What was being tested in the simulation run?
• What parameters were changed?
• What transaction data (if any) was changed?
• What were the performance implications due to the change?
• What were the business benefits or negatives associated with the change? (in order to make the simulation blog as scientific as possible, both negative as well as positive outcomes or simulations runs must be documented.
• What were the implications for different areas of the supply chain (i.e. manufacturing vs. distribution, or regional distribution center vs. lower level distribution centers)? This relates to a main point that simulation changes should not be evaluated in isolation, but the overall impacts must be documented. This is why simulation is a difficult and time-consuming task.
• Was the change migrated to the live system? If not, why not, if so, what were the results in the live environment.

All of these items can be placed in a form on the blog allowing them to easily searched or even extracted to a spreadsheet for comparison.

WordPress, a blogging platform, provides a high degree of organization to information. It solves the problem of websites pre-blogging, in that they lacked a natural organization, and had to have organization placed upon them. Finally, WordPress or similar software can be installed on company servers and have the same security level as other corporate html pages. This information management is perfect for documenting and archiving successive simulation runs.

Developing Simulation Prioritization

The important thing is that the previous simulation results are captured and are that what was done and the results can be understood to that simulations do not need to be constantly re-run. Eventually, through documentation, heuristics or rules of thumb will be built up that are based upon the previously run simulations. This can help prioritize future simulation runs. If one parameter alteration has been run 4 times, but a different parameter alteration has never been simulated, an argument can be made to give the untested parameter priority.

How Prevalent is Simulation?

We performed a literature review on simulation specifically for SAP APO-SCM. What I found is that there is not very much written on SAP SCM simulation, and this brings up the question of how commonly companies are actually performing simulation. With the requirements regarding the support of simulation, it is clear that this is an area that could easily drop between the cracks. In my personal consulting experience I have seem companies spend more time talking about simulation than doing it.

Simulation and S&OP

What I am observing is that S&OP, which is primarily a simulation environment in any case, is gathering the majority of emphasis in terms of corporate priority. My concern is that S&OP will become the singular simulation environment for planning. This would be unfortunate because the types of things that S&OP looks for, are different that the more functional areas want to verify. For instance, S&OP would not be interested in checking for the changes in lot size.

However, what would make sense is to combine all simulation requirements related to supply chain into one area, and prioritize requests from S&OP, Supply Planning, Demand Planning, Production Planning, Transportation Planning, etc. in one queue.

Conclusion

Simulation a very beneficial capability to have. SAP SCM is not an ideal simulation environment, but for those companies that run APO-SCM in production, it will mostly likely be the simulation environment as well. Simulation can be greatly enhanced through the application of a consistent approach in terms of what is documented and documenting the results in either a blogging platform or other content management system. How prevalent simulation is in SCM is very much an open issue, and because it is not frequently performed, there is not very much literature or distributed experience as to how to do it, much less do it well. Companies who do sign up for simulation in SCM must be cognizant of the amount of work they are signing up for. As I have pointed out in this post, there is quite a bit to consider, and not all of it simply software related, to creating an effective simulation capability.

References

http://www.scdigest.com/assets/FirstThoughts/07-05-31.php?cid=1073&ctype=content

Supply Chain Management on Demand: Strategies, Technologies Applications, Chae An and Steve Buckley

Real Optimization with SAP APO, Josef Kallrath and Thomas I. Maindl

Supply Chain Simulation Modeling Made Easy: An Innovative Approach, Cope, Fayez, Mollanghasemi, Kaylani

More notes on simulation

Simulation modeling is a versatile and powerful tool that has grown in popularity due to its ability to deal with complicated models of corresponding complicated system (Kelton, Sadowski, & Sadowski, 2002; Wartha et al., 2002). Nevertheless, simulation models can be time consuming to build, requiring substantial development time, effort and experience. According to Mackulak, Lawrence & Colvin (1998), simulation development time takes about 45% of the total simulation project effort. Furthermore, simulation-modeling efforts often have to be modified to accommodate the development of what if scenarios and constantly changing requirements. These modifications also take time to model. An alternative to creating a unique simulation model is to reuse an existing generic model that can be reconfigured for individual projects. - Supply Chain Simulation Modeling Made Easy: An Innovative Approach