Judith and Charles were standing in the front yard of their Tiki Island home when a sharp boom pulsed through, rattling their windows and skipping their family truck a few inches off the ground. Five miles away and a few hours earlier, an isomerization unit was restarting after a turnaround, setting off a chain of events that would go on to become one of the worst oil refinery disasters in recent history, with a death toll of 15.
The March 23, 2005 Texas City disaster reminds the industry that the costs of inefficient processes and communications issues — two common problems plaguing turnarounds— can become more than financial for oil and gas companies.The good news is that some companies are rapidly figuring out the solution to these issues in their IT department, namely in mobile technology and data analytics. In a time of depressed oil prices and a shaky recovery, the need to revamp the refinery turnaround process is more imperative than ever.
The main pain points with implementing these solutions lie in the legacy systems and huge amount of technical debt in place. The IT and OT convergence of these solutions requires a skillful and careful modernization of the legacy system.
Before the Turnaround: Strategic and Detailed Planning Phases
Refinery turnarounds don’t fail in the execution phase; they fail in the planning phases. What dooms these phases are the assumptions that planners make around assets, processes and people. Without data, and by that we mean real, objective numbers from the field, some planners are left to reviewing manuals or other other materials to determine the flow and structure of the work. Some planners do have access to the data, but their company has not made the subsequent investment in the user experience of manipulating that data for it to be of any use to the team.
For assets, knowing the condition and crucial manufacturing information can drive a much more efficient prioritization process. Age can only tell so much about a product. Sensor data can not only provide the condition and performance, but can go a step farther with advanced analytics by providing self-diagnosis of issues. Preventative maintenance, or maintenance that relies on age only to determine the maintenance schedule, is only effective for assets that have an established time-based failure pattern. Although this philosophy does work for the 18 percent of assets in most plants, it is cost-prohibitive. Parts are replaced prematurely and a larger inventory is needed for regular replacements, as well as causing increased labor costs for regular replacements.
The advantage of connecting the sensor data to mobile applications is the increase in efficiency for the remote workforce. For one of our remote monitoring applications, the team dramatically increased their productivity and job satisfaction. Not having to lug a laptop around, even during social events when on-call, meant the team could still respond to issues without powering up a laptop and could also quickly manipulate the data to diagnose the issue. Read more about it in the case study.
Information on the people and processes hinges on the ability of the planning team to have data from field observations. Most employees rely on workarounds to become more efficient, especially if their day involves ineffective software. The management team makes assumptions of how they think work is getting done out in the field, and then passes that to the planning team. The reality is often very different. For example, one large industrial company assumed that the inspections of sensitive materials were dragging down the productivity of the operations, and sought out ways to make the inspections team faster. Once our team of researchers went out into the field, it was readily apparent that communications issues were causing the backlog. The inspections were done in record time, but the materials sat idle as the approvals process stretched longer and longer. Solving these communications issues saved millions of dollars.
When there are 30,000 separate procedures going on during a refinery turnaround, those assumptions quickly add up. If time budgeting for the workforce is even 5 percent off, that’s the equivalent of adding another 1,500 procedures to the turnaround. Enterprise mobile applications can assist with the planning phases here by tracking work orders and workforce time very closely. Pairing that with an investment in field observations from human factors researchers can go a long way toward eliminating all assumptions.
During the Turnaround: The Execution Phase
The plant has ceased operations. With careful and knowledgable planning, everything should run on-schedule and on-budget.
Where does this go wrong?
Even the most diligent planning cannot make up for the communications issues that are inherent to the large temporary workforce. Poor communication and management visibility lend themselves to efficiency issues going unnoticed and undiagnosed. There are issues that teams can solve during the turnaround; too often, management does not hear about them until the debrief when it is too late.
Communications issues are not the fault of the team. It’s the paper processes and manual approvals that drag this down. While we can’t point out specific inefficiencies for each procedure that goes on during the turnaround, here are our general best practices to solve these issues:
- Invest in real-time, step-by-step communication with the workforce– that works for the workforce. Don’t cut the end-user out of your development process for a digital product. IT, business and the end-user need to all have input into what ends up out in the field. It’s important here to make sure that the product is developed to work within the context of the user. For example, any application with a white background is going to be difficult to see in the harsh sunlight of the day. Asking users to pinch and zoom is also problematic if they are going to be holding another tool in their hand. These are insights you get out of ethnographic research of the end-users.
- Don’t overwhelm users. Often, companies load up on training and safety materials for employees all at once. Once they get out in the field though, this information can be difficult to recall. Using geofencing and contextually aware technologies, companies can deliver the crucial health and safety procedures to employees at the exact time needed, instead of relying on the employee to review procedures before going onto the work site. Communicating this effective information right before the employee works with hazardous equipment will dramatically decrease the incidents on plants. A landmark psychology study showed that for memory, people remember the first (primacy effect) and last items (recency effect) the best, and the middle portions in between the worst. Bombarding your employees with a whole series of steps before they do a crucial process is in direct opposition to this research, but is often necessary if the entire process is not fully mobilized. An Apple Watch or other wearable technology enables smaller chunks of information at the exact moment needed to maximize the primacy and recency effects, and see better adherence to crucial safety procedures.
- Build real-time dashboards to keep management in the loop. Management don’t need to see every single data point. Often, they just need the high-level view of turnaround progress. Real-time dashboards with a powerful back-end can make this a reality for management. Oil and gas companies need to do away with the paper weekly reports that can show issues that are already well on their way to contributing to a cost overrun. Real-time data on high-level metrics can help the team manage by exception and deal with issues as they happen.
After the Turnaround: The Debrief
For most companies, by the time the debrief happens, the planning for the next turnaround is already underway. This does not mean that debriefs are not worthwhile exercises. Debriefing with all of the stakeholders on the team, and digging into their insights with data, can yield very valuable information to improve with turnaround. Invest in another round of research to prioritize improvements for the next release.