Hydropower plant flow control
From mountain lake to hydropower plant
By Westrow Cooper
Overlooking Dravadals-lake at about 1000m above sea-level. The lake is fed by the Folgefonna glacier and is one of three reservoirs for the Jukla power plant. The water conduit between the lake and the power plant is about 6.6km onto which the two other lakes connect, and it contains an human made cavern within the mountain and that acts as an air cushion chamber. Credit: Flow Design Bureau
Norway’s mountain lakes offer a stunningly beautiful landscape – and a major resource for hydropower. But how to manage and protect the water flow down to the plant and optimize use of hydropower for a sustainable energy future?
These are the questions being addressed by Morten Kjeldsen and colleagues at Flow Design Bureau in Stavanger, Norway. We had an opportunity to speak with Morten at a recent conference where he described how the team is carrying out edge analysis using an executable digital twin – or what they prefer to call a ‘virtual sensor.’
“At present we’re calling it a virtual sensor,” said Morten, “though in many ways it is a subset of a digital twin because you need a good model to actually establish that sensor.” The team uses a virtual sensor rather than a regular physical sensor due to the specific circumstances of hydropower in Norway:
“The water is drawn from lakes, acting as reservoirs, high up in the mountains, and so must travel through many kilometers of piping, tunnels and conduits before its final run into the penstock and turbine at the power plant. The difference in elevation between the mountains and the turbine means there can be anything from 200 to 1200 meters difference in head. Furthermore, the long tunnels create huge inertia, making it challenging to switch efficiently between hydropower and, say, power from wind farms when the wind is blowing.”
Creating a ‘natural’ battery
In providing edge analysis the team helps to optimize energy management so that water drawn when there is an excess of wind power can for some plants be pumped back more efficiently and without causing harm up to the lake, ready for use when the wind is low. In effect, the system acts as a large battery, but a battery that is charged naturally. Demand is expected to grow significantly in the coming years so there are plans to build much larger storage plants and associated infrastructure.
This means Norway that can adopt a greener, more sustainable solution for energy generation, without having to rely on gas turbines to take up the slack when, for example, the wind isn’t blowing. However, the most common hydro turbine unit in Norway is a fixed design, which allows only minor adjustments to the amount of water flowing into the turbine. They were constructed to be kept running to provide a constant output, not to be switched up and down in response to varying environmental conditions for other sources of power.
Through digitalization the team is helping to extend the value and useful life of the existing infrastructure by enabling continued operation of existing units to meet today’s changing requirements: “Switching up or decreasing the power output from the hydro-plant creates mass flow oscillations in the long tunnels carrying the water,” Morten explained. “This can threaten to wear away or even destroy the tunnels because they are carved out of rock. Equally, the water can carry rock and sediments towards the turbine units, which is obviously not something you want.”
Insight over the entire length of flow
This is where the virtual sensors really come into their own. Placing instruments in the actual tunnels is very difficult because they are very long and high up in the mountains. Yet it is essential to know what is happening in terms of, say, velocity, especially in the surge shafts where there are frequent and significant changes in water level.
“The virtual sensor provides a dynamic model of the power plant,” Morten continued. “We combine this with the expected flow rate of the water running through the turbine unit and actual measurements of levels, where possible. With those inputs and an accurate model, we can identify the fluctuations in level and flow rate oscillations in the tunnel and display them in real time. The operator thus has insight into what’s happening over the entire length of the tunnels. This delivers huge value for the plant operator.”
Water accidentally exiting valve/gate-house due to mass-surges in the water conduit and because of unfavourable operation of turbine units in the power plant. Picture by Statkraft. Used in a presentation by Kvæstad and Kjeldsen at PTK 2016. PTK is an annual Norwegian hydropower conference and exhibition by Renewables Norway.
Building the model with Simcenter™ Flomaster™
When asked about software, Morten replied: “Our favorite software for building the model is Simcenter Flomaster. In addition to using Simcenter Flomaster for the principal model, we also create a simplified, dynamic model, implemented directly in the code, that we use for data acquisition – the virtual sensors pick up information that was, previously, largely unavailable and unknown.
“Validating the model is not possible in a conventional sense, but by measuring different inputs against each other we can check the accuracy of the model. So, for example, the virtual sensors in the water conduit are checked against the control action of the turbine unit, as well as pressure transducers just upstream of the turbine unit. If we manage to calculate the value correctly using the model, then we can deduce the rest of the calculated performance is correct.”
Using Artificial Intelligence to predict fluctuations
Although the Flow Design Bureau team does not use the term, the use of sensors providing valuable information, edge computing, real time data and cloud computing together add up to what many engineers would call a digital twin.
At present the team is delivering this service to one operator in Norway, and they have a third party who applies machine learning on the data that is provided. Because the team collects time series as historical data, in future they aim to be able to better predict the optimum compromise between rapid response to changing power demands and allowable flow rates in the water conduits. AI can also process the engineering data from the sensors together with written notes from technicians on site to provide additional levels of insight.
Gaining new perspectives
Morten sees many advantages in operating as a small consultancy, both personally and for the industry as a whole: “I think the beauty of small companies like ours is that we can be hands-on and contribute to all aspects of an installation, from placing the sensors in the correct position on site to installing and upgrading the IT equipment. So – proper engineering and not just sitting in an office and writing models!”
“Sometimes we engineers can spend too much time in the office in front of a computer or, conversely, always in overalls on site, but not both. I think it’s important to get out and meet colleagues from all parts of the profession, to share ideas and gain new perspectives on a subject. That’s where to find inspiration and be really creative.”
Click here for more information on Simcenter Flomaster
“With those inputs and an accurate model, we can identify the fluctuations in level and flow rate oscillations in the tunnel and display them in real time. The operator thus has insight into what’s happening over the entire length of the tunnels.”