How well is the digital twin idea conceived, and implemented in oil & gas industry?
Anyone having the experience in actually implementing, and looking at the benefits /challenges?
Digital Twin Idea
Re: Digital Twin Idea
Just to start with, there has been a lot of initiative on digitizing the assets, their functional data, assessment, day-to-day analysis and among all those, this big giant, Digital Twin is also the one. We were give to evaluate this task with one of the service supplier and we tried working out as how a twin would first be developed. Took us just 6 months on scoping that what is needed to be included as this was supposed to be an assessment task and at the same time should be representative enough that if cleared, could be expanded to other assets. A unit was chosen and we started gathering the data. I would not be commenting on the data assessment and analysis here as all that was heavily depending on the quality and quantum of the data collected. We got nominations from all disciplines to participate with us and while we were gathering the data, we realized that the current system itself was having quite a number of challenges:
1. We got to know that the personnel operating & maintaining our plant themselves do not have everything known of the assets (knowledge & experience transfer came out to be the first gap)
2. Way the data was being maintained got hybrid controls, something in hard copy formats being handled separately and a lot many things having soft copies only and a lot of documents yet to be converted to have their soft version (though initially of less use)
3. 40 years plant was not having all 40 years of its data available in the right form (confidence foundation of digital twin)
1. We got to know that the personnel operating & maintaining our plant themselves do not have everything known of the assets (knowledge & experience transfer came out to be the first gap)
2. Way the data was being maintained got hybrid controls, something in hard copy formats being handled separately and a lot many things having soft copies only and a lot of documents yet to be converted to have their soft version (though initially of less use)
3. 40 years plant was not having all 40 years of its data available in the right form (confidence foundation of digital twin)
Re: Digital Twin Idea
Not directly related to digital twin however I lately had an experience in assessing a digital HAZOP tool which can help going through the HAZOP Workshop with digitized P&IDs, fed into a model picking up the applicable scenarios, provding existing mitigations and supporting with risk ranking along with the recommendations, if applicable.
Issue is the solution provides a lot more support but there is still a requirement that the team members conduct a workshop and assess all scenarios and applicable safeguards.
And then consider such solutions only for support purposes may not be economically viable.
Issue is the solution provides a lot more support but there is still a requirement that the team members conduct a workshop and assess all scenarios and applicable safeguards.
And then consider such solutions only for support purposes may not be economically viable.
Re: Digital Twin Idea
Having all the similar thoughts on digital HAZOP.
though i liked the idea that modeling does the basework but the issue of reviewing everything all over again. what i suggested to the solution provider was to enhance the modeling of the backend enabling the users to rely on the specific inputs but generic models do not help.
though i liked the idea that modeling does the basework but the issue of reviewing everything all over again. what i suggested to the solution provider was to enhance the modeling of the backend enabling the users to rely on the specific inputs but generic models do not help.
ivani1 wrote: 22 Jul 2025, 04:50 Not directly related to digital twin however I lately had an experience in assessing a digital HAZOP tool which can help going through the HAZOP Workshop with digitized P&IDs, fed into a model picking up the applicable scenarios, provding existing mitigations and supporting with risk ranking along with the recommendations, if applicable.
Issue is the solution provides a lot more support but there is still a requirement that the team members conduct a workshop and assess all scenarios and applicable safeguards.
And then consider such solutions only for support purposes may not be economically viable.
Re: Digital Twin Idea
Actually, we recently completed modeling which came up with the following challenges:
1. Data collection:
It was a complex process to collect the data. System teams expect the technical teams to actually take the complete lead as they play a support role only. Before coming to this step of collecting the data, expectations should be made clear to the technical team that what eventually a twin would be able to do. Data collection should start after. What we did most of the times was to interact with the system teams, understand their needs to build models, and then they asking us back what exactly we need. Now data itself stands challenging when someone asks you that how many process changes have already been made and what kind of impact we had. Objective here is to build trending based upon historical data but not for everything, we have everything. The most available data at all times was with the process, integrity, and inspection teams however, maintenance and operations were lacking the trail on most of the items.
I would strongly recommend here to have separate collaboration meetings within all technical disciplines before engaging with any internal or external system team. They know a lot but nothing beyond then you know about your processing facility. Mandatory technical disciplines include operations to start with (any good looking model would look acceptable to any other team but if that does not have the language of what operations use to communicate with the plant, it will remain unacceptable and of no real use).
2. Skilled Workforce & Awareness:
You build on what you have available as the basic skills with your resources engaged. We asked for maintenance data, historical failures, RCAs conducted, studies, outcome, recommendations, implementation, and then we learnt that this is a lot to ask from people not sufficiently trained to not only produce all this but enable the system team to integrate all this in the models. And then after having models as well, only the skilled staff would make adoption useful. What benefit you would expect if the workforce interacting with the twin is not skilled.
Before engaging any resource here, you make sure that he has got an idea of what is expected with this initiative and what support you would be requiring.
Implementing and maintaining high-fidelity digital twin models can be challenging.
3. Data Privacy
And time & again, you would have to run to your IT team to allow engaged system team use the provided data, and then terms & conditions. You better know how slow this goes and sometimes it does not move for months, tagging the issue, "a critical cybersecurity matter".
1. Data collection:
It was a complex process to collect the data. System teams expect the technical teams to actually take the complete lead as they play a support role only. Before coming to this step of collecting the data, expectations should be made clear to the technical team that what eventually a twin would be able to do. Data collection should start after. What we did most of the times was to interact with the system teams, understand their needs to build models, and then they asking us back what exactly we need. Now data itself stands challenging when someone asks you that how many process changes have already been made and what kind of impact we had. Objective here is to build trending based upon historical data but not for everything, we have everything. The most available data at all times was with the process, integrity, and inspection teams however, maintenance and operations were lacking the trail on most of the items.
I would strongly recommend here to have separate collaboration meetings within all technical disciplines before engaging with any internal or external system team. They know a lot but nothing beyond then you know about your processing facility. Mandatory technical disciplines include operations to start with (any good looking model would look acceptable to any other team but if that does not have the language of what operations use to communicate with the plant, it will remain unacceptable and of no real use).
2. Skilled Workforce & Awareness:
You build on what you have available as the basic skills with your resources engaged. We asked for maintenance data, historical failures, RCAs conducted, studies, outcome, recommendations, implementation, and then we learnt that this is a lot to ask from people not sufficiently trained to not only produce all this but enable the system team to integrate all this in the models. And then after having models as well, only the skilled staff would make adoption useful. What benefit you would expect if the workforce interacting with the twin is not skilled.
Before engaging any resource here, you make sure that he has got an idea of what is expected with this initiative and what support you would be requiring.
Implementing and maintaining high-fidelity digital twin models can be challenging.
3. Data Privacy
And time & again, you would have to run to your IT team to allow engaged system team use the provided data, and then terms & conditions. You better know how slow this goes and sometimes it does not move for months, tagging the issue, "a critical cybersecurity matter".
Re: Digital Twin Idea
Thanks neo.
neo wrote: 27 Jul 2025, 13:21 Actually, we recently completed modeling which came up with the following challenges:
1. Data collection:
It was a complex process to collect the data. System teams expect the technical teams to actually take the complete lead as they play a support role only. Before coming to this step of collecting the data, expectations should be made clear to the technical team that what eventually a twin would be able to do. Data collection should start after. What we did most of the times was to interact with the system teams, understand their needs to build models, and then they asking us back what exactly we need. Now data itself stands challenging when someone asks you that how many process changes have already been made and what kind of impact we had. Objective here is to build trending based upon historical data but not for everything, we have everything. The most available data at all times was with the process, integrity, and inspection teams however, maintenance and operations were lacking the trail on most of the items.
I would strongly recommend here to have separate collaboration meetings within all technical disciplines before engaging with any internal or external system team. They know a lot but nothing beyond then you know about your processing facility. Mandatory technical disciplines include operations to start with (any good looking model would look acceptable to any other team but if that does not have the language of what operations use to communicate with the plant, it will remain unacceptable and of no real use).
2. Skilled Workforce & Awareness:
You build on what you have available as the basic skills with your resources engaged. We asked for maintenance data, historical failures, RCAs conducted, studies, outcome, recommendations, implementation, and then we learnt that this is a lot to ask from people not sufficiently trained to not only produce all this but enable the system team to integrate all this in the models. And then after having models as well, only the skilled staff would make adoption useful. What benefit you would expect if the workforce interacting with the twin is not skilled.
Before engaging any resource here, you make sure that he has got an idea of what is expected with this initiative and what support you would be requiring.
Implementing and maintaining high-fidelity digital twin models can be challenging.
3. Data Privacy
And time & again, you would have to run to your IT team to allow engaged system team use the provided data, and then terms & conditions. You better know how slow this goes and sometimes it does not move for months, tagging the issue, "a critical cybersecurity matter".