All Member Open Forum

Hello all, 

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts. 

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there. 

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Hi Brayden, I have a hyper free-text search engine that works very well for analyzing free text. It works extremely well for identifying failures using work order description and technician repair comments. When failures becomes your main criterion of focus, everything else falls into place.

Cheers,

------------------------------
Jeff Wahl
Maintenance Manager, CMMS
Greenbank WA
------------------------------

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Jeff, what is this engine and where can I get one?!

---------------------------------
Karl Burnett

Solvay, Inc
Anderson SC
---------------------------------



Original Message:
Sent: 01-14-2021
From: Jeff Wahl
Subject: RE: Similar Bad Actors from SAP

Hi Brayden, I have a hyper free-text search engine that works very well for analyzing free text. It works extremely well for identifying failures using work order description and technician repair comments. When failures becomes your main criterion of focus, everything else falls into place.

Cheers,

------------------------------
Jeff Wahl
Maintenance Manager, CMMS
Greenbank WA
------------------------------

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Hi Karl,
I call it Contextual Text Analysis. I had my first iteration running in SQL, then brought it to VBA in Excel. My third iteration fine tuned separated CTA into a stand-alone search engine running in Excel. This has been my strong belief and passion ever since I got involved in Reliability. Over the years I have added several features to make my analysis easier, faster, and more accurate.

I've studied free-text fields in CMMS(es), operations log, spare parts consolidations, operational excellence improvement schemes, and even PLC error coding. All I need is to have free-text fields that can be queried out of the system.

If you have free-text field data, I can show you how to load it and set up your search parameters. Remove any identifying information beforehand. If you keep generic equipment names, dates and other useful categorizing data, I can show you all you want. Let me know. I might not be on this site after my membership dues comes up. Feel free to check out my IMC-2015 presentation on my LinkedIn home page: linkedin.com/in/jeffwahl.

Best regards,

------------------------------
Jeff Wahl
Maintenance Manager, CMMS
Greenbank WA
------------------------------

Original Message:
Sent: 01-14-2021 01:01 PM
From: Karl Burnett
Subject: Similar Bad Actors from SAP

Jeff, what is this engine and where can I get one?!

---------------------------------
Karl Burnett

Solvay, Inc
Anderson SC

Original Message:
Sent: 01-14-2021
From: Jeff Wahl
Subject: RE: Similar Bad Actors from SAP

Hi Brayden, I have a hyper free-text search engine that works very well for analyzing free text. It works extremely well for identifying failures using work order description and technician repair comments. When failures becomes your main criterion of focus, everything else falls into place.

Cheers,

------------------------------
Jeff Wahl
Maintenance Manager, CMMS
Greenbank WA

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Brayden,

Two things I'd recommend. First, if your work management process and coding of work orders allows, I would see if you can determine which of the corrective work orders were unplanned/unscheduled work orders. If that is implied in the definition of Priority 1 and 2, then disregard this one. A proper work management process should allow you to differentiate between them. Second, I suggest that you incorporate operational, EHS and quality costs into your bad actor identification method. Incorporating these costs will make it much more likely that your efforts at improvement will get support from those managers and from the plant manager. Never work against your peers, it's a team that should be focused on the mission. Including peer manager concerns will foster better communication and teamwork. Influencing changes to process and procedures that you don't directly control are easier when functions work together. 

Best regards,
Tom

------------------------------
Tom Moriarty
President, Alidade MER, Inc.
Satellite Beach FL
http://www.alidade-mer.com
(321) 773-3356
tjmpe@alidade-mer.com
------------------------------

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Hi Tom

Thanks for your reply! To clarify, I'm using the P1 and P2 codes to identify unplanned work on a grand scale. Unfortunately, it's not the perfect filter for this and I've found myself manually weeding through the data after the initial export. This is the best way that I've been able to filter the type of work that I'm looking for since our plant has never used any other sort of code in SAP to identify reactive work directly. (In the past I've seen specific work order types for reactive-corrective maintenance in Maximo - this allowed supervisors to bypass procurement, so in a way it was beneficial for them as well as for tracking. Unfortunately, it's never been setup in SAP here).

As I mentioned to Suzy below, While it may come as a surprise, production losses are very few and far between at our site, in fact, since commissioning Canaport LNG has never missed a gas nomination from our pipeline partners. As such, tracking actual production losses in the conventional sense is not a factor that I've given much consideration to. On the other hand, using our equipment availability (and potentially factoring in equipment criticality from a risk perspective) may be another way of capturing this variable.

Thanks again for your thoughts, Tom!

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Original Message:
Sent: 01-15-2021 09:31 AM
From: Tom Moriarty
Subject: Similar Bad Actors from SAP

Brayden,

Two things I'd recommend. First, if your work management process and coding of work orders allows, I would see if you can determine which of the corrective work orders were unplanned/unscheduled work orders. If that is implied in the definition of Priority 1 and 2, then disregard this one. A proper work management process should allow you to differentiate between them. Second, I suggest that you incorporate operational, EHS and quality costs into your bad actor identification method. Incorporating these costs will make it much more likely that your efforts at improvement will get support from those managers and from the plant manager. Never work against your peers, it's a team that should be focused on the mission. Including peer manager concerns will foster better communication and teamwork. Influencing changes to process and procedures that you don't directly control are easier when functions work together.

Best regards,
Tom

------------------------------
Tom Moriarty
President, Alidade MER, Inc.
Satellite Beach FL
http://www.alidade-mer.com
(321) 773-3356
tjmpe@alidade-mer.com

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Hello Brayden,

Try running IW69 - Failure Analysis. This is out of the box SAP functionality which holds huge potential for categorizing your bad actors. Look at your default layout and expand on it if you wish. There are so many more fields that have value that you can pull in. You can see WONUM, Notification, F/L, EQ, Priority, Malfunction start/stop, breakdown,,,etc. As is true with any asset driven application, the data your retrieve is only as good as the data entered over time. So what is key, is training. Train your staff to populate your Notifications per your business process especially damage & cause codes. Encourage entering supplemental text. Your pivot table will love this.

------------------------------
Ed Espinosa, CMRP, CRL, PMP
Sr. Performance Analyst
Puget Sound Energy
Bellingham WA
------------------------------

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Contextual Failure Analysis works by searching all keywords and key phrases simultaneously. Keyword "hits" are recorded on the row in which it was discovered.

I usually create a draft set of keywords that I think will bring out the data, either by experience or by interviewing someone knowledgeable of the failure. It is during this time that I start to gain understanding of the data, and the natural language wording used by the people at the particular site. This process always gives me insights for new keywords to add, which brings out data that I might want to filter in or out of my list. I usually make 6 or 7 passes during this process, while at the same time, making notes in my failure notes. Re-running Keyword Search does not affect my notes so each pass improves my results.

To help you visualize how this works, I put together a few example keyword lists for various problems I have studied. Notice the variations of what seem to be the same word or acronym. Humans are not very consistent when it comes to data entry... I can say this from my own personal experience.

Cheers,

------------------------------
Jeff Wahl
Maintenance Manager, CMMS
Greenbank WA
------------------------------

Original Message:
Sent: 01-15-2021 12:35 PM
From: Edward Espinosa
Subject: Similar Bad Actors from SAP

Hello Brayden,

Try running IW69 - Failure Analysis. This is out of the box SAP functionality which holds huge potential for categorizing your bad actors. Look at your default layout and expand on it if you wish. There are so many more fields that have value that you can pull in. You can see WONUM, Notification, F/L, EQ, Priority, Malfunction start/stop, breakdown,,,etc. As is true with any asset driven application, the data your retrieve is only as good as the data entered over time. So what is key, is training. Train your staff to populate your Notifications per your business process especially damage & cause codes. Encourage entering supplemental text. Your pivot table will love this.

------------------------------
Ed Espinosa, CMRP, CRL, PMP
Sr. Performance Analyst
Puget Sound Energy
Bellingham WA

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Hi Jeff, 

Thanks for all of your replies! The concept of contextual failure analysis seems like a great tool for digging into the various failure modes of an asset It seems like it would definitely help to grab some of the work orders that are misplaced in the "General" or "Other" cause codes. It may also be helpful for grouping similar failures across multiple redundant assets.

I can't help but think there should be an easier way to do this using standard SAP functionality - not the text analysis, but the grouping of similar sub-equipment failures.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Original Message:
Sent: 01-16-2021 04:38 PM
From: Jeff Wahl
Subject: Similar Bad Actors from SAP

Contextual Failure Analysis works by searching all keywords and key phrases simultaneously. Keyword "hits" are recorded on the row in which it was discovered.

I usually create a draft set of keywords that I think will bring out the data, either by experience or by interviewing someone knowledgeable of the failure. It is during this time that I start to gain understanding of the data, and the natural language wording used by the people at the particular site. This process always gives me insights for new keywords to add, which brings out data that I might want to filter in or out of my list. I usually make 6 or 7 passes during this process, while at the same time, making notes in my failure notes. Re-running Keyword Search does not affect my notes so each pass improves my results.

To help you visualize how this works, I put together a few example keyword lists for various problems I have studied. Notice the variations of what seem to be the same word or acronym. Humans are not very consistent when it comes to data entry... I can say this from my own personal experience.

Cheers,

------------------------------
Jeff Wahl
Maintenance Manager, CMMS
Greenbank WA

Original Message:
Sent: 01-15-2021 12:35 PM
From: Edward Espinosa
Subject: Similar Bad Actors from SAP

Hello Brayden,

Try running IW69 - Failure Analysis. This is out of the box SAP functionality which holds huge potential for categorizing your bad actors. Look at your default layout and expand on it if you wish. There are so many more fields that have value that you can pull in. You can see WONUM, Notification, F/L, EQ, Priority, Malfunction start/stop, breakdown,,,etc. As is true with any asset driven application, the data your retrieve is only as good as the data entered over time. So what is key, is training. Train your staff to populate your Notifications per your business process especially damage & cause codes. Encourage entering supplemental text. Your pivot table will love this.

------------------------------
Ed Espinosa, CMRP, CRL, PMP
Sr. Performance Analyst
Puget Sound Energy
Bellingham WA

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Hey Brayden - In the case of CMMS functionality, "easy" translates to generic compatibility. That is also the main reason failure code don't work. I worked at a company for 13 years where we were regulated by the FDA under the strictest of rules. We set up failure codes with contextual dropdown menus to help reduce the clutter when closing work orders. That didn't work. 

With Contextual Failure Analysis, a "finely tuned" group of failure keywords/ key phrases are populated on only the work order rows of focus. One of my most effective improvements was to design "relevance" into the results. That is to say, with the click of a button, the software counts and sorts the results from highest to lowest "keyword hit count". It is an amazing feeling to see only those work order lines relevant to your search sort right to the top of 12,000 lines. CFA cuts through the clutter quickly - nothing else I've seen does this so effectively. 

One more bonus of your finely tunes group of keywords/ key phrases: You can now run the keywords against your work orders daily (or as often as you like), to look for specific failures in near-real-time. For costly or intermittent failures this is a great way to monitor failures - by letting the peoples' own natural language as your guide. I urge you to try it. You sound very capable of running a SQL query on a year's worth of work orders. I can show you step-by-step how quickly and effectively this works.

The next move is yours. I've talked myself out.

Cheers,

------------------------------
Jeff Wahl
Maintenance Manager, CMMS
Greenbank WA
------------------------------

Original Message:
Sent: 01-18-2021 07:40 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hi Jeff,

Thanks for all of your replies! The concept of contextual failure analysis seems like a great tool for digging into the various failure modes of an asset It seems like it would definitely help to grab some of the work orders that are misplaced in the "General" or "Other" cause codes. It may also be helpful for grouping similar failures across multiple redundant assets.

I can't help but think there should be an easier way to do this using standard SAP functionality - not the text analysis, but the grouping of similar sub-equipment failures.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com

Original Message:
Sent: 01-16-2021 04:38 PM
From: Jeff Wahl
Subject: Similar Bad Actors from SAP

Contextual Failure Analysis works by searching all keywords and key phrases simultaneously. Keyword "hits" are recorded on the row in which it was discovered.

I usually create a draft set of keywords that I think will bring out the data, either by experience or by interviewing someone knowledgeable of the failure. It is during this time that I start to gain understanding of the data, and the natural language wording used by the people at the particular site. This process always gives me insights for new keywords to add, which brings out data that I might want to filter in or out of my list. I usually make 6 or 7 passes during this process, while at the same time, making notes in my failure notes. Re-running Keyword Search does not affect my notes so each pass improves my results.

To help you visualize how this works, I put together a few example keyword lists for various problems I have studied. Notice the variations of what seem to be the same word or acronym. Humans are not very consistent when it comes to data entry... I can say this from my own personal experience.

Cheers,

------------------------------
Jeff Wahl
Maintenance Manager, CMMS
Greenbank WA

Original Message:
Sent: 01-15-2021 12:35 PM
From: Edward Espinosa
Subject: Similar Bad Actors from SAP

Hello Brayden,

Try running IW69 - Failure Analysis. This is out of the box SAP functionality which holds huge potential for categorizing your bad actors. Look at your default layout and expand on it if you wish. There are so many more fields that have value that you can pull in. You can see WONUM, Notification, F/L, EQ, Priority, Malfunction start/stop, breakdown,,,etc. As is true with any asset driven application, the data your retrieve is only as good as the data entered over time. So what is key, is training. Train your staff to populate your Notifications per your business process especially damage & cause codes. Encourage entering supplemental text. Your pivot table will love this.

------------------------------
Ed Espinosa, CMRP, CRL, PMP
Sr. Performance Analyst
Puget Sound Energy
Bellingham WA

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Hi Edward,  

Thanks for your reply! I've used IW69 in the past for further analysis of bad actors, but never as the initial data generation code. It's been a recent effort of mine to review and update our cause/damage catalogues. In 2018, >40% of our corrective maintenance orders had a damage code of "General" or "Other" - a big reason for this was that the generic list of all cause codes was applied to each equipment type. This made the process seem like just another useless task for supervisors, especially when you could make a selection such as "Hard Drive Failure" for something like a water recirculation pump. 

Junk in = Junk out. Thanks again, Edward.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Original Message:
Sent: 01-15-2021 12:35 PM
From: Edward Espinosa
Subject: Similar Bad Actors from SAP

Hello Brayden,

Try running IW69 - Failure Analysis. This is out of the box SAP functionality which holds huge potential for categorizing your bad actors. Look at your default layout and expand on it if you wish. There are so many more fields that have value that you can pull in. You can see WONUM, Notification, F/L, EQ, Priority, Malfunction start/stop, breakdown,,,etc. As is true with any asset driven application, the data your retrieve is only as good as the data entered over time. So what is key, is training. Train your staff to populate your Notifications per your business process especially damage & cause codes. Encourage entering supplemental text. Your pivot table will love this.

------------------------------
Ed Espinosa, CMRP, CRL, PMP
Sr. Performance Analyst
Puget Sound Energy
Bellingham WA

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Edward,
As a person who knows how to work with data, I once in a large refinery tried to ferret out bad actors from the CMMS data--defined as failures that repeat within a few month of the same failure. It was a given, but never proven, feeling that there were a tremendous number of bad actors in the plant. It was extremely, almost impossible, to tease bad actors out of the data in SAP. I found that it takes active daily involvement to do what cannot be completely automated, so much that your plant likely does not experience enough such cases to make it worthwhile. Furthermore, the bad actors that would justify attention usually make themselves known to everyone--they are obvious.

What I've observed over the years is that the idea of bad actors is a sexy notion. Consequently, like the refinery, it is on faith that if we reversed them, we would make a huge contribution to uptime and maintenance expense. 

Final comment. Be careful about running down that rabbit hole.

------------------------------
Richard Lamb
Analytics4strategy.com
rchrd.lamb@gmail.com Houston TX
http://analytics4strategy.com
------------------------------

Original Message:
Sent: 01-18-2021 07:32 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hi Edward,

Thanks for your reply! I've used IW69 in the past for further analysis of bad actors, but never as the initial data generation code. It's been a recent effort of mine to review and update our cause/damage catalogues. In 2018, >40% of our corrective maintenance orders had a damage code of "General" or "Other" - a big reason for this was that the generic list of all cause codes was applied to each equipment type. This made the process seem like just another useless task for supervisors, especially when you could make a selection such as "Hard Drive Failure" for something like a water recirculation pump.

Junk in = Junk out. Thanks again, Edward.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com

Original Message:
Sent: 01-15-2021 12:35 PM
From: Edward Espinosa
Subject: Similar Bad Actors from SAP

Hello Brayden,

Try running IW69 - Failure Analysis. This is out of the box SAP functionality which holds huge potential for categorizing your bad actors. Look at your default layout and expand on it if you wish. There are so many more fields that have value that you can pull in. You can see WONUM, Notification, F/L, EQ, Priority, Malfunction start/stop, breakdown,,,etc. As is true with any asset driven application, the data your retrieve is only as good as the data entered over time. So what is key, is training. Train your staff to populate your Notifications per your business process especially damage & cause codes. Encourage entering supplemental text. Your pivot table will love this.

------------------------------
Ed Espinosa, CMRP, CRL, PMP
Sr. Performance Analyst
Puget Sound Energy
Bellingham WA

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Brayden,

For LNG bad actors list application, I would think production loss cost is the primary criterion than maintenance cost.  I want to echo Tom Moriarty's advice.  You can work with the Production team to access their production database where production losses are captured by asset (tagnumbers).  You can still include maintenance cost in the criteria just to track where the annual maintenance cost went.  I would guess your top 10 bad actors are mostly equipment that cause plant shutdowns incurring production losses.

Another program to identify reliability improvement opportunity is to focus on repeat failures. In CMMS, search what equipment with priority 1 & 2 as you said and the most frequent corrective WO's that may not cause production loss, but equipment repair (redundant system or opportunity window).  Getting WO's information for these repeat failure modes may present an opportunity to investigate further and warrant a RCFA (root cause failure analysis). 

I hope this helps!

------------------------------
SUZY JIANG
Staff Reliability Engineer
Sugar Land TX
------------------------------

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Hi Suzy,

Thanks for your reply! While it may come as a surprise, production losses are very few and far between at our site, in fact, since commissioning Canaport LNG has never missed a gas nomination from our pipeline partners. As such, tracking actual production losses in the conventional sense is not a factor that I've given much consideration to. On the other hand, using our equipment availability (and potentially factoring in equipment criticality from a risk perspective) may be another way of capturing this variable. 

I have also included a count of the work orders over the same time periods in my pivot table. I guess my question would be, what is the appropriate way to allocate investigation resources when you begin to use several methodologies to identify bad actors? The reason that I've used Pareto analysis on our reactive maintenance costs is so that I can investigate the equipment that causes 80% of those costs, but if I'm also generating another list based on the number of failures, what takes precedence when it comes to assigning the work?

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Original Message:
Sent: 01-16-2021 04:05 PM
From: SUZY JIANG
Subject: Similar Bad Actors from SAP

Brayden,

For LNG bad actors list application, I would think production loss cost is the primary criterion than maintenance cost.  I want to echo Tom Moriarty's advice.  You can work with the Production team to access their production database where production losses are captured by asset (tagnumbers).  You can still include maintenance cost in the criteria just to track where the annual maintenance cost went.  I would guess your top 10 bad actors are mostly equipment that cause plant shutdowns incurring production losses.

Another program to identify reliability improvement opportunity is to focus on repeat failures. In CMMS, search what equipment with priority 1 & 2 as you said and the most frequent corrective WO's that may not cause production loss, but equipment repair (redundant system or opportunity window).  Getting WO's information for these repeat failure modes may present an opportunity to investigate further and warrant a RCFA (root cause failure analysis).

I hope this helps!

------------------------------
SUZY JIANG
Staff Reliability Engineer
Sugar Land TX

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Hi Breyden,
Looks like you got lots of good insight from Jeff/Ed/Suzy/etc... I agree with many of their comments about trying to leverage SAP's existing coding ability, looking at production loss vs. mtce opex, and then utilizing a more sophisticated tool like Jeff's vs. some simple find() functions in excel.  

A couple additional comments:
1. Another tool out there that does some pretty solid text analysis and doesn't require a degree in data science is JMP Pro - we've used their text explorer on WO/notif short text and long text - it quickly finds the common words/phrases, allows for stemming (e.g. pump* to catch pump, pumps, pumping, ...) and then does a very good job grouping similar terms/phrases.  Something to think about.  I've built some excel and SQL-based tools but the JMP one is very slick out of the box.  Having said that, I would still love to see Jeff's tool in action.

2. The SAP coding is a great step but isn't one that will yield immediate benefits - think of it as setting up your successor for success.  We've been actively pushing this initiative at some of our sites lately and it's so nice to have structured codes for bad actor codes (e.g. Code 1234 = Mechanical Seal) vs. having to rely on the MANY ways folks will spell it in the WO or notification.  For us, we found that one big reason why folks weren't coding was that they didn't see the value (i.e. reliability didn't prove its worth) and the existing coding options were way too big/silly (e.g. our electrical equipment would have 500+ codes to pick from).  We've trimmed that down to reasonable numbers (~8-25 per equipment type) to catch most of the pains in the butt so hopefully in the coming years we have much more structured 'bad actor' data than having to sift thru the text.   Our data science folks are also working on a Machine Learning / NLP tool that they hope will be able to 'auto-code' the work as well --> for now we're going to keep manually coding and use our results to teach their tool so it can hopefully replace the manual effort once it proves to be very accurate.

Hope that helps and good luck.

------------------------------
Benjamin Horne PEng
Reliability Engineer
Imperial Oil
Calgary AB
------------------------------

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Original Message:
Sent: 1/20/2021 10:50:00 PM
From: Benjamin Horne
Subject: RE: Similar Bad Actors from SAP

Hi Breyden,
Looks like you got lots of good insight from Jeff/Ed/Suzy/etc... I agree with many of their comments about trying to leverage SAP's existing coding ability, looking at production loss vs. mtce opex, and then utilizing a more sophisticated tool like Jeff's vs. some simple find() functions in excel.  

A couple additional comments:
1. Another tool out there that does some pretty solid text analysis and doesn't require a degree in data science is JMP Pro - we've used their text explorer on WO/notif short text and long text - it quickly finds the common words/phrases, allows for stemming (e.g. pump* to catch pump, pumps, pumping, ...) and then does a very good job grouping similar terms/phrases.  Something to think about.  I've built some excel and SQL-based tools but the JMP one is very slick out of the box.  Having said that, I would still love to see Jeff's tool in action.

2. The SAP coding is a great step but isn't one that will yield immediate benefits - think of it as setting up your successor for success.  We've been actively pushing this initiative at some of our sites lately and it's so nice to have structured codes for bad actor codes (e.g. Code 1234 = Mechanical Seal) vs. having to rely on the MANY ways folks will spell it in the WO or notification.  For us, we found that one big reason why folks weren't coding was that they didn't see the value (i.e. reliability didn't prove its worth) and the existing coding options were way too big/silly (e.g. our electrical equipment would have 500+ codes to pick from).  We've trimmed that down to reasonable numbers (~8-25 per equipment type) to catch most of the pains in the butt so hopefully in the coming years we have much more structured 'bad actor' data than having to sift thru the text.   Our data science folks are also working on a Machine Learning / NLP tool that they hope will be able to 'auto-code' the work as well --> for now we're going to keep manually coding and use our results to teach their tool so it can hopefully replace the manual effort once it proves to be very accurate.

Hope that helps and good luck.

------------------------------
Benjamin Horne PEng
Reliability Engineer
Imperial Oil
Calgary AB
------------------------------

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Hi Richard,
Yes, it has evolved a lot in recent years.  I too used it in university mainly for its 'traditional stats' capabilities, which are also VERY useful for data-minded RE's - they have add-ins for DMAIC/6sigma and reliability so that all the core stuff of Weibull analysis, SPC charts, etc.... are readily available.  I find it very powerful (as compared to say Minitab) for the stats work, but Minitab seems easier and with a better UI for that kind of traditional stats work IMHO anyway.
Text explorer was added in the last few-is years to JMP Pro and I don't know off hand what the cost is.  It's not cheap, but nor is it insane like some geoscience stuff that costs more than a car.  

There are likely purpose built tools that are better specifically in the text mining space, but JMP Pro is very good in many areas.  One additional feature in its text explorer that I forgot to mention is that not only can it find common words (pump, motor, leak, ...) and phrases (cracked pipe, leaking seal, ...) but it can also cluster common words that are often found together in the same cell but not back-to-back (e.g. 'seal' and 'leak' might be common friends even if they don't appear back to back and in the same order each time) which is VERY handy ('the seal is leaking...', 'we found a leaking seal....', 'we replaced a seal that ops found to be leaking...') would all be likely found in the cluster analysis.  

Hope that helps.  Cheers,
Ben.

------------------------------
Benjamin Horne PEng
Reliability Engineer
Imperial Oil
Calgary AB
------------------------------

Original Message:
Sent: 01-21-2021 07:40 AM
From: Richard Lamb
Subject: Similar Bad Actors from SAP

Benjamin,

It sounds as if JMP Pro is onto something. I used it for a time in grad school but it seems to have added some text mining capabilities. What is the license price to the software?

Richard G. Lamb, PE, CPA

Original Message:
Sent: 1/20/2021 10:50:00 PM
From: Benjamin Horne
Subject: RE: Similar Bad Actors from SAP

Hi Breyden,
Looks like you got lots of good insight from Jeff/Ed/Suzy/etc... I agree with many of their comments about trying to leverage SAP's existing coding ability, looking at production loss vs. mtce opex, and then utilizing a more sophisticated tool like Jeff's vs. some simple find() functions in excel.

A couple additional comments:
1. Another tool out there that does some pretty solid text analysis and doesn't require a degree in data science is JMP Pro - we've used their text explorer on WO/notif short text and long text - it quickly finds the common words/phrases, allows for stemming (e.g. pump* to catch pump, pumps, pumping, ...) and then does a very good job grouping similar terms/phrases.  Something to think about.  I've built some excel and SQL-based tools but the JMP one is very slick out of the box.  Having said that, I would still love to see Jeff's tool in action.

2. The SAP coding is a great step but isn't one that will yield immediate benefits - think of it as setting up your successor for success.  We've been actively pushing this initiative at some of our sites lately and it's so nice to have structured codes for bad actor codes (e.g. Code 1234 = Mechanical Seal) vs. having to rely on the MANY ways folks will spell it in the WO or notification.  For us, we found that one big reason why folks weren't coding was that they didn't see the value (i.e. reliability didn't prove its worth) and the existing coding options were way too big/silly (e.g. our electrical equipment would have 500+ codes to pick from).  We've trimmed that down to reasonable numbers (~8-25 per equipment type) to catch most of the pains in the butt so hopefully in the coming years we have much more structured 'bad actor' data than having to sift thru the text.   Our data science folks are also working on a Machine Learning / NLP tool that they hope will be able to 'auto-code' the work as well --> for now we're going to keep manually coding and use our results to teach their tool so it can hopefully replace the manual effort once it proves to be very accurate.

Hope that helps and good luck.

------------------------------
Benjamin Horne PEng
Reliability Engineer
Imperial Oil
Calgary AB

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Original Message:
Sent: 1/21/2021 9:37:00 AM
From: Benjamin Horne
Subject: RE: Similar Bad Actors from SAP

Hi Richard,
Yes, it has evolved a lot in recent years.  I too used it in university mainly for its 'traditional stats' capabilities, which are also VERY useful for data-minded RE's - they have add-ins for DMAIC/6sigma and reliability so that all the core stuff of Weibull analysis, SPC charts, etc.... are readily available.  I find it very powerful (as compared to say Minitab) for the stats work, but Minitab seems easier and with a better UI for that kind of traditional stats work IMHO anyway.
Text explorer was added in the last few-is years to JMP Pro and I don't know off hand what the cost is.  It's not cheap, but nor is it insane like some geoscience stuff that costs more than a car.  

There are likely purpose built tools that are better specifically in the text mining space, but JMP Pro is very good in many areas.  One additional feature in its text explorer that I forgot to mention is that not only can it find common words (pump, motor, leak, ...) and phrases (cracked pipe, leaking seal, ...) but it can also cluster common words that are often found together in the same cell but not back-to-back (e.g. 'seal' and 'leak' might be common friends even if they don't appear back to back and in the same order each time) which is VERY handy ('the seal is leaking...', 'we found a leaking seal....', 'we replaced a seal that ops found to be leaking...') would all be likely found in the cluster analysis.  

Hope that helps.  Cheers,
Ben.

------------------------------
Benjamin Horne PEng
Reliability Engineer
Imperial Oil
Calgary AB
------------------------------

Original Message:
Sent: 01-21-2021 07:40 AM
From: Richard Lamb
Subject: Similar Bad Actors from SAP

Benjamin,

It sounds as if JMP Pro is onto something. I used it for a time in grad school but it seems to have added some text mining capabilities. What is the license price to the software?

Richard G. Lamb, PE, CPA

Original Message:
Sent: 1/20/2021 10:50:00 PM
From: Benjamin Horne
Subject: RE: Similar Bad Actors from SAP

Hi Breyden,
Looks like you got lots of good insight from Jeff/Ed/Suzy/etc... I agree with many of their comments about trying to leverage SAP's existing coding ability, looking at production loss vs. mtce opex, and then utilizing a more sophisticated tool like Jeff's vs. some simple find() functions in excel.

A couple additional comments:
1. Another tool out there that does some pretty solid text analysis and doesn't require a degree in data science is JMP Pro - we've used their text explorer on WO/notif short text and long text - it quickly finds the common words/phrases, allows for stemming (e.g. pump* to catch pump, pumps, pumping, ...) and then does a very good job grouping similar terms/phrases.  Something to think about.  I've built some excel and SQL-based tools but the JMP one is very slick out of the box.  Having said that, I would still love to see Jeff's tool in action.

2. The SAP coding is a great step but isn't one that will yield immediate benefits - think of it as setting up your successor for success.  We've been actively pushing this initiative at some of our sites lately and it's so nice to have structured codes for bad actor codes (e.g. Code 1234 = Mechanical Seal) vs. having to rely on the MANY ways folks will spell it in the WO or notification.  For us, we found that one big reason why folks weren't coding was that they didn't see the value (i.e. reliability didn't prove its worth) and the existing coding options were way too big/silly (e.g. our electrical equipment would have 500+ codes to pick from).  We've trimmed that down to reasonable numbers (~8-25 per equipment type) to catch most of the pains in the butt so hopefully in the coming years we have much more structured 'bad actor' data than having to sift thru the text.   Our data science folks are also working on a Machine Learning / NLP tool that they hope will be able to 'auto-code' the work as well --> for now we're going to keep manually coding and use our results to teach their tool so it can hopefully replace the manual effort once it proves to be very accurate.

Hope that helps and good luck.

------------------------------
Benjamin Horne PEng
Reliability Engineer
Imperial Oil
Calgary AB

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

We've set up a criteria for bad actor in terms of
-total cost of repair incurred $ 20000 &
- 2 or more failures on the same equipment  
This help us to to retrieve the list of equipment's from SAP. Then the SME's along with reliability engineer will do the RCFA and come up with a action plan. Once the action plan is implement, the equipment is monitored for year and pulled out from the bad actor list, provided no failure reoccur. The equipment still remain on tracking list for another two years    

Just for your information, we've over 100000 asset tag's in SAP and out of which 75,000 are "A&B"  critical and remaining is "C" critical in our 550 MBD refinery.

------------------------------
Francis Castelino PE
Rotating Equipment Engineering Specialist
ARAMCO
Rastanura
------------------------------

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Hi Francis, 

Thanks for your feedback. That sounds like a very similar system to ours. Do you generate the Bad Actors list on an annual basis? 

Do you take any steps to identify and combine common failure modes between similar equipment? (i.e. on a bank of process cooling fans, we recognized that each of the motors on VFD's were experiencing relatively low MTBF and had common failure modes of fluting from electrical discharge on both bearing races. This was easily detectible by the maintenance staff, but I'm still hoping to use our SAP data to generate these types of cases - I think as I talk myself through it, the best way for us to do this might still be to simply rely on our cause codes).

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------

Original Message:
Sent: 01-29-2021 08:20 AM
From: Francis Castelino
Subject: Similar Bad Actors from SAP

We've set up a criteria for bad actor in terms of
-total cost of repair incurred $ 20000 &
- 2 or more failures on the same equipment
This help us to to retrieve the list of equipment's from SAP. Then the SME's along with reliability engineer will do the RCFA and come up with a action plan. Once the action plan is implement, the equipment is monitored for year and pulled out from the bad actor list, provided no failure reoccur. The equipment still remain on tracking list for another two years

Just for your information, we've over 100000 asset tag's in SAP and out of which 75,000 are "A&B"  critical and remaining is "C" critical in our 550 MBD refinery.

------------------------------
Francis Castelino PE
Rotating Equipment Engineering Specialist
ARAMCO
Rastanura

Original Message:
Sent: 01-14-2021 07:26 AM
From: Brayden Olafson
Subject: Similar Bad Actors from SAP

Hello all,

Today I'm seeking some knowledge and experience relating to SAP, and in particular how to use it to effectively generate data for a Bad Actors list/program.

To date, I've presented our bad actors list as a function of Pareto analysis on the annual reactive maintenance costs of each of our individual assets. The way that I've achieved this is by using IW38 to generate a list of all WO2 (Corrective work orders) with a priority of 1 or 2 (Very High, or High). I use these parameters to get as close to "Reactive" as  possible, since our maintenance department has never taken steps to categorize them otherwise. For the most part, these parameters are quite accurate in filtering out the appropriate "Reactive" work, however some manual filtering can be neccessary.

I then group the costs of each asset in a Pivot table, and calculate the cumulative costs starting at the most costly reactive equipment. The top 80% of our costs are typically attributed to a range of 5%-12% of our 420 SAP managed assets. For each of these assets, I perform a quick pareto analysis on their failure modes in order to generate a backlog for our continuous improvement/RCA efforts.

One thing that I would like to expand this effort to is being able to group similar assets and their failure modes so that I can identify if certain similar assets are, as a whole, impacting our budget on a similar scale to some individual bad actors. I've tried to think of a way to use our functional location and/or equipment tags to do this, however, I can't help but think there is a more efficient method out there.

Any thoughts or suggestions are appreciated.

------------------------------
Brayden Olafson, CMRP
Reliability Specialist, Canaport LNG, Saint John, NB
brayden.olafson@canaportlng.com
------------------------------