Photo: Lene Esthave

Preview of tomorrow’s control room

Production and management

For each litre of oil extracted from the North Sea, 10 litres of water must be pumped into the underground.    

A research project will help Total optimize the operation of the complex plants that clean enormous amounts of sea water and pump the water into the ground under high pressure.

Imagine a control room where an alarm has gone of, so that all employees are on their toes. Then remove the sound. That is the atmosphere at the open-plan office on the fourth floor of Total’s building in the port of Esbjerg.

“Alarms are always ringing up here,” says Project Engineer Steven Munk Østergaard Lauridsen of Total, but emphasises that these alarms have no effect on the safety on the platforms.

Critical alarms are handled by a completely different system, ESD (Emergency Shutdown), which triggers an automatic, safe shutdown. Fortunately, this happens extremely rarely.

“The alarms we deal with here are a matter of technical irregularities that have to be addressed to ensure that the plants are operated optimally. Oil production systems are complex, and technical irregularities both large and small constantly crop up”.

Steven Munk Østergaard Lauridsen’s field of work is water injection.

In newspaper graphics, you often see offshore recovery illustrated as a straw stuck in a black bubble below the seabed. But the times where the pressure of the North Sea reservoirs was sufficiently high to automatically push oil into the pipelines are long gone.

Today, operators constantly pump enormous amounts of water down into the compact layers of lime to drive the oil out of hiding.

Water injection is now the subject of Total's part of a project led by the Danish Hydrocarbon Research and Technology Centre at DTU.

Based on models developed by Professor Emeritus Morten Lind’s group at DTU Electrical Engineering, the parties of the project develop software to help the operators in control rooms distinguish between critical and less critical technical alarms.

In addition, the system must inform the operator of the underlying cause of the individual alarm.

The importance of stable water injection

Back when they started to pump water into the underground, only a small amount was needed to achieve an effect. Over time, the amount of water had to be increased.

"The project is a major step towards further digitization. "
Steven Munk Østergaard Lauridsen

As a rule of thumb, 10 litres of water are currently needed to extract one litre of oil. In total, this amounts to more than 400,000 cubic metres of water a day. There is certainly plenty of water available in the North Sea, but you cannot simply use sea water.

The system consists of seven modules. In each module, the water goes through a form of either mechanical, chemical, or biological purification before it is pumped into the underground under very high pressure—300 bar.

“Today, we have a strong focus on water treatment, and even though the safety risks aren’t as great as when handling oil or gas, the technical complexity is just as high. This means there will be just as many reports of technical irregularities,” explains Steven Munk Østergaard Lauridsen.

At the same time, the financial impact is high.

“If the water injection has to be suspended due to a breakdown, it will heavily impact oil recovery.”

This also means there are major financial gains by ensuring stable water injection.

Cracks in the limestone must be avoided

Put another way, financially it makes sense for Total to optimize water injection.

“If you increase the water pressure, the production of oil should increase. But if the pressure becomes too high, the limestone layer cracks. Cracks act as tiny motorways for the water. No matter how much water you pump down, most will flow through the cracks. The oil will sit in the areas of limestone that are never touched by water. In other words, you have partially destroyed your reservoir,” explains Steven Munk Østergaard Lauridsen.

It’s a balancing act to find the correct pressure.

“You want to find the ‘sweet spot’. Meaning the maximum pressure where you don’t make cracks in the reservoir. But when you don't know where this point is exactly, it is necessary to keep a very large safety margin and thus lose production.”

This is where the question of alarms enters the picture.

If the pressure drops on a particular valve, it may be a specific technical problem or an operating error. But it can also be a sign that something is wrong in the reservoir—for example that you are approaching the point where cracks form.

An optimized alarm system will present the possible causes of the pressure drop and also the probability of whether these are correct. This will give you a better basis for taking action.

At the same time, you can run close to the optimum pressure, confident that you will get certain types of alarms before crossing the line.

Eliminating pointless alarms

Today, the basic principle is to start by defining the normal range of operations. That is, the acceptable intervals for the most important parameters such as pressure, temperature, flow, particle content, etc. If the parameter approaches the edge of its interval, an alarm is triggered.

“The problem is that at this point there is very little time for the operator to try less extensive interventions to see if they have the expected effect. The smart thing would be to exploit the patterns in the constant movements within the normal range to see if the system is getting out of balance—before it’s actually out of balance,” explains Steven Munk Østergaard Lauridsen.

“In the future, we hope that operators can avoid having to make as many decisions and that they will get support, making it easier for them to make the right decisions.”

Research linked to reality

For the past 25 years, a group headed by Morten Lind at DTU Electrical Engineering has researched ways to improve systems for control rooms. This has created the ‘Multilevel Flow Modelling’ method which sits at the core of this project.

Working ‘multilevel’ means that the system looks for the causes for a specific alarm at several levels.

In some cases, it is simply a technical error. In other cases, an operator may have changed a setting without thinking it could have a detrimental effect elsewhere in the system. And finally, it may be down to poor management decisions.

“This idea is very useful for us. We had already taken the initiative to work with ‘alarm management’ where we are working to reduce the number of meaningless or outright unnecessary alarms,” says Steven Munk Østergaard Lauridsen.

In addition to the Danish Hydrocarbon Research and Technology Centre at DTU and DTU Electrical Engineering, DTU Compute and the Department of Energy Technology at Aalborg University are academic partners in the project.

On the business side, Total is joined by ConocoPhillips UK and the young Norwegian company Eldor Technology in order to develop and market the commercial end product. This will be under the trademark AlarmTracker.

“The project shows the value of the Danish Hydrocarbon Research and Technology Centre at DTU,” says Steven Munk Østergaard Lauridsen.

“If we, as a company, approach a research institution with only a research purpose in mind, it usually ends in a PhD dissertation. This may be okay sometimes, but there’s often a need for someone to make sure that the research results are linked to reality. As an energy company, we are not the ones to develop products such as AlarmTracker. That is why it’s so nice to have the Centre at the head of things, so we end up with a product we can use.”

Big data to exploit back catalogues of alarms

“As a general rule, AlarmTracker is installed offshore to help operators. But I imagine that we will initially let the system run as a digital twin to the current alarm system. After some time, we can evaluate the system. Does it detect all the critical issues as it should? Does it give the right underlying explanations for the alarms?” says Steven Munk Østergaard Lauridsen.

The long-term expectations are high:

“The project is a major step towards further digitization. It opens the door for getting even more value from the large amounts of data we generate about our production system.”

Since 2000, the company has stored all alarms, and since 2012, data from the sensor readings has also been added. The dream is to create an intelligent system that will benefit from this enormous data catalogue. This means that when you get a certain type of alarm today, the system can identify whether the situation is a complete match to a critical situation that might have occurred four years ago and state what the reason was back then—and whether the solution worked.

Steven Munk Østergaard Lauridsen summarizes:

“We can’t avoid the alarms. But we can hopefully avoid that the performance of a single person on a given day determines how wisely we act on them.”