Riyadh-headquartered utility and green hydrogen developer and investor Acwa Power has topped MEED’s annual power and water developer ranking over the past few years.

The company’s portfolio, which it values at approximately $94bn, includes 49 thermal and renewable energy power plants and about a dozen water desalination plants.

These assets can generate 65GW of power and 8 million cubic metres a day of desalinated water.

Acwa Power continues to bid and win new contracts at home and abroad under Marco Arcelli, who was appointed as the firm’s chief executive shortly after Paddy Padmanathan, the firm’s CEO for 18 years, stepped down in March last year.

The company has tapped artificial intelligence (AI) and the stacks of technologies behind it to enable its future strategy, notes Thomas Altmann, the company’s executive vice-president for innovation and new technology.

“At Acwa Power, we are not talking about AI, we are doing it,” says Altmann, who cites that the company has developed an in-house algorithm to enable an augmented AI or human-in-the-loop (HITL) AI application.

Through this application, a plant operator may receive data or advice from an AI-enabled module that can trigger a response after the data is validated through the operator’s experience.

“We are focusing on human-in-the-loop, under the umbrella of collective intelligence … this in-house artificial neural network (ANN) algorithm has so far contributed to about 12% cost reduction of chemical dosing in one of our water desalination plants, which encourages us to industrialise this technology and implement it in selected plants,” explains Altmann.

“I think generally that AI is not stoppable; we’re not using it as a buzzword; we are focusing very much on use cases that make sense and create bottom-line impact.”

Analytics and machine learning

Acwa Power has been at the forefront of innovations not just in operating its plants but in winning tenders by proposing the use of new technologies.

“Over the past decade, we managed to reduce power consumption in our desalination plants by over 80%,” notes Altmann.

“I recall in 2005 when Acwa Power submitted the first bid, most of the desalination plants that were built during that period were based on thermal water desalination technologies such as MSF.”

However, things changed when Saudi Arabia tendered the Shuqaiq 2 independent water and power project (IWPP), which, for the first time, did not prescribe a specific technology for the project’s desalination unit. 

There are a lot of things that typically are not discussed, like having to write new procedures to distinguish tasks that can be done by robots, algorithms and humans

“This allowed Acwa Power to innovate and deploy for the first time a membrane-based desalination technology at scale in Saudi Arabia. We were the only consortium that offered to build a 100% reverse osmosis plant in combination with a power plant (IWPP). Our bid was successful by offering 17% to 19% lower tariff due to significantly lower energy consumption compared to an MSF plant,” Altmann said.

Several years later, with the Rabigh 3 independent water project, the offtaker specified a drastically reduced energy consumption. Altmann said they “had to press the reset button, turn every stone and create a paradigm shift in RO design to meet these requirements”.

At this point, Acwa Power has used the so-called Typical Meteorological Year (TMY) methodology for renewable energy to predict future power generation in solar plants.

Altmann subsequently introduced a similar Typical Seawater Year (TSY) methodology, which used the previous five years’ worth of seawater data, used big data analytics to understand seawater resources, and designed their plant according to this result.

“This contributed significantly to our successful bid because we used real data rather than assumed data based on the request for proposals and implemented several design improvements, which resulted in the lowest ever specific energy consumption for RO in the region,” the executive noted.

Altmann says Acwa Power also introduced the so-called pressure centre in an RO plant in Saudi Arabia, where a high-pressure pump in a desalination plant is not necessarily linked to one reverse osmosis (RO) train. Instead, a pipe connects the pump and the racks, and each pipe can fit any rack. This allows fewer and larger pumps to be used and improves efficiency.

“Rabigh 3 was a breakthrough, and we continued to further optimise the process, and the results were applied, for example, in Taweelah in Abu Dhabi and Jubail 3A in Saudi Arabia. 

“We added a solar component to the Taweelah IWP as an innovation and we continued to fine-tune and optimise as we move forward.”

Altmann also says they were the first to introduce machine learning to reduce chemical costs and predict the optimal time for membrane cleaning in RO desalination plants in the region.

The goal is to continue innovating into the future, says Altmann, citing their research and development (R&D) centres in leading universities across the GCC, in particular at the King Abdullah University of Science and Technology in Saudi Arabia, where they operate centres of excellence focusing on water, solar, hydrogen and AI.

AI and the future of utility jobs

While a fully autonomous water desalination plant may still be a few years away depending on how fast AI technologies develop, Altmann acknowledges that future plants will have fewer people on the floor.

This does not necessarily mean large-scale staff displacement since “we keep winning new plants, and we can reassign and retrain or reskill our staff.”

“As some jobs disappear, new jobs will be created,” adds Altmann. “There are many opportunities to utilise experienced people.”

The executive, however, cautions that AI deployment in a company is not just a matter of installing software codes.

It requires a change in culture and processes, particularly in HR, where one has to move away from thinking of employees’ positions or jobs but rather their tasks.

“One needs to distinguish which tasks require a lot of data, and involve routines, and which can be done by an algorithm, versus tasks involving creativity, human interaction or validation against ethical standards or privacy compliance and so on.”

“There are a lot of things that typically are not discussed, like having to write new procedures to distinguish tasks that robots, algorithms and humans can do,” he continues.

The executive also cites the paramount importance of the quality of data and the AI readiness of the Internet of Things (IoT) system to enable AI applications.

“The most important thing, besides ethics and privacy, from a technical perspective, is data. If you want a high-quality prediction or an advisory module, you need to put most of your effort into the data first.

“Utilising an algorithm … that’s the easy part, the difficult one is to get clean data, eliminate bias, noise and spurious correlations and consider differential shifts in the training of data since AI works differently to a human brain. AI doesn’t have an intuition or awareness to sense biases and is susceptible to providing wrong predictions,” explains Altmann.

Renewable-powered desalination plants

Altmann argues that if green hydrogen can be produced using 100% renewable energy, the same can be applied to water production.

Acwa Power built a 20MW solar PV to complement the grid-sourced electricity supply for the Taweelah IWP in Abu Dhabi because, according to Altmann, the RFP did not disallow it.

They are looking at doing more of these projects where it makes sense from a sustainability and efficiency point of view.

“The (Taweelah) RFP did not disallow the installation of a solar PV, and there was an available space, so we went ahead to build an on-site solar PV farm, which allowed us to reduce more expensive energy import from the grid.”

Altmann asserts that desalinated water can have a zero-carbon footprint by building captive water desalination plants or sourcing clean energy from the grid.

However, moving to a 100% renewable source will increase the complexity of building desalination plants.

“There’s a difference from a technical perspective if you take power from the grid, which has certain stability and inertia and is often linked to power plants.

“You will need to redesign the desalination plants differently, with a different operation strategy and different motors, and to deploy long-duration energy storage due to intermittency of renewable energy,” he concludes.