Professor Ilić featured in Portugal's Público newspaper


February 18, 2015

Electrical and Computer Engineering and Engineering and Public Policy Professor Marija Ilić was recently featured in Portugal's Público. Her expertise in smart energy networks is highlighted in the English translation below:

We can't produce renewable energy if we don't use it wisely

She has spent the last four years studying ways to introduce smart electricity distribution systems in the Azores network in cooperation with the Green Islands project, launched in 2009 by the Massachusetts Institute of Technology. She has already written a book with the results.

Marija Ilic, 63, was born in the former Yugoslavia but lives in the US, where she specializes in intelligent energy networks. As a researcher at Carnegie Melon University (CMU) she coordinates the Dynamic Monitoring and Decision Systems (DyMonDS) project – design and software implementations of intelligent system for electric power grids. She has developed partnership with scientists and Portuguese universities, under the CMU Portugal Program, funded by the Foundation for Science and Technology. EDP and EDA have already shown interest in adopting it.

Marija Ilic, a specialist in smart energy networks, argues that governments should not set a goal for renewable energies without first knowing in detail what the consumers need.

For the Green Islands Project you’ve studied the energy networks, and the demand characteristics in the Azores, especially in Flores and São Miguel. What problems did you find?
In São Miguel there is an old power station in the middle of the island that uses expensive and polluting fuel. We wanted to see whether it would be possible to use less of this fuel, and install wind and solar parks to provide clean energy to consumers. The technical problem was: how do you control the entire system so that, even when there is no wind, you can provide the cleanest energy possible? In Flores there is a mini-hydro in which you can control the energy produced by the dam, but the capacity of this plant may not be sufficient to control all wind power variations without careful scheduling of other resources and storage. The energy from the conventional power station is also easily controllable, but with the wind and the sun it is not like that. However, it is possible to predict wind, with a day or an hour in advance, and then schedule the distribution of energy in order to supply the predicted amount.

The weather forecast is critical?
Yes, very. The more accurate is the forecast, the smaller will be the differences between what the power station produces and what the consumer needs. Because wind energy has to be stored quickly, it is very expensive. The better we can forecast and schedule what we can control, the smaller requirements are for expensive storage needed to balance the supply to customers when the wind does not blow.

What tools are needed?
Basically algorithms and computer software. We must forecast and optimize the amount of energy whose supply can be scheduled based on this prediction.

This is only possible with wind or also with other resources?
The wind was our concern because it is the largest resource in São Miguel and Flores. But solar energy is also a concern. In the book “The Tale of Two Low-Cost Green Azores Islands” we approach first what happens if we apply these technologies only to conventional energies. After that we compared what happens if we schedule two variables: wind power and demand. In this case we propose Adaptive Load Management (ALM), contrary to what is currently done by the power companies. They use direct storage control in which there is a charge and the customer agrees to occasional interruptions of supply. With ALM, the consumer says, with a day or a week in advance that he wants to use a certain amount of power and is willing to pay for it. This is a consumer’s choice and with this information the company controls the supply of energy.

Is this system already working?
So far we have only simulated the two islands. The next step is to develop the software in the control center (the Azores Electricity) in São Miguel and also on the side of the consumers to help adjust to predictable wind power components.  Moreover, it is critical to install fast automation to control frequency and voltage deviations by controllable plants and fast storage, like flywheels, to cancel the effects of hard-to-predict deviations from predicted power production. Information technologies will help to balance the system without necessarily using traditional power plants. And so it is possible to reduce the consumer’s energy bills and pollution. When we talk about information exchange, consumers are concerned about security issues. But it is very safe because the client does not have to say exactly which loads he has at home, he just says how much energy he needs, in what range.

Imagine that the consumer makes a prediction for a week, and then he needs more than that. What happens?
That's a good question. For example, if you have a refrigerator you cannot commit to a day in advance, because the refrigerator cools and warms up in an hour. So it only makes sense that consumers tell us if refrigerators are cool or warm  only one hour in advance. With air conditioners, that don’t exist in large numbers in the islands, a wider prediction in time is possible. The idea is that the consumer complies with what he has undertaken. The system is based on an interactive exchange of information and trust that the different parts will do what they say. From time to time you can do an audit, but how we can strengthen this trust is still an open question.

Is it possible to leave the Azores "laboratory"?
What we did in the Azores is not exclusive to the islands. We are also talking to the EDP’s and EDA’s representatives. The model on which we are working, designated DyMonDS, is ready to be applied. Basically we want to have a model that, in a given power distribution system, can analyze the characteristics of the customers and available resources, and decide what information is necessary to exchange, at what level, so that the system is more sustainable. In our research we raised a question: would electric vehicles work in the islands? The answer was yes. We concluded that if they are equipped with  intelligent charging which responds  directly to  weather forecasting similar to other electricity users, they could over time, contribute to lowering long-term  energy costs on the islands. Basically, the user does not charge the vehicle only when he must ; he plans to charge it at an optimal time  in advance.

Won’t it be difficult for a consumer to predict when he has to charge the vehicle?
Everything is based on the correct information exchange. This can’t be done directly by the consumer. There will be software installed in the house’s electrical system which enables and disables the thermostat, taking into account the signals from  the operator. If the operator  sends a signal to say that there will be a lot of wind, the smart charging software embedded within the vehicle adjusts and decides the vehicle must be charged within 24 hours in order to achieve lower costs and allow the balance of the power system. This is the vehicles intelligent charging, as opposed to rapid charging which is where everyone charges the battery at the same time and there is a power surge. The CMU project aims to understand how to incorporate DyMonDS software in the housing system for purposes of aligning customer preferences with what the system can provide.

It will be expensive?
Not at all. The software is inexpensive once developed. The challenge is to understand what the consumer preferences are. Many researchers are developing this type of software but what distinguishes us is that we know the physics of the islands and we have the knowledge to integrate the software in the network. DyMonDS uses an integrated technology both in the system of network users and also in the exchange of information that must be made between the distributor and the users so that they work together.

The existing network remains?
Yes, the basic infrastructure is the same but will be equipped with smart devices.

Electric vehicles seem to have a great importance in this process, but currently there are very few in Portugal, about 900. Do you believe that this could change in a short term?
It is very likely, on the continent. There are signs encouraging electric vehicles. There are electric buses in public transport service, for example. If there is an economic incentive  signal to the cars because they don’t pollute, this will lead to higher penetration of electric vehicles.

The Government currently gives an incentive to purchase.
But it is a subsidy. We shouldn’t subsidize, but adopt a system that will work always. You can start with the subsidy but then it compensates the us of these vehicles in the long term. And if there are long-term contracts between consumers and energy companies, there is a commitment and investors will invest in these vehicles. It won’t pay off in a day. But it is like this everywhere, electric cars are penetrating slowly.

Is EDP interested in your model?
They have shown great interest. As there are more resources distributed by the system, such as wind, solar or micro generated, it becomes more difficult to define which information should be exchanged, and how to coordinate the clients, how to balance the system. The mechanism they have today to make such control is called SCADA (supervisory control and data acquisition), which circulates the information from the operator to the substation. But below that no one knows what happens. What we suggest is that there should be more exchange of information, in an interactive way, incorporated to facilitate system management. They are very interested in this and have called our DyMonDS "the new SCADA".

In the book you argue that no one can commit to a certain renewable energy production capacity without understanding first the characteristics of the existing resources and consumer preferences. But it is not what most governments have done?
Yes, absolutely. For environmental reasons it was good to have only renewable resources, but depending on when they can generate and how much the production costs, it is not obvious that we have to replace the existing power plants with pre-determined percent of renewables. There is not enough caution in the definition of renewable goals, people don’t think about what is best for the system considering what consumers want and are willing to pay, how much it costs, and what are the environmental benefits.

In the case of the Azores, the government wants to have 75% of renewable energy by 2020.
Yes. In the second phase of this project we want to explore that issue if f the Government really wants to have 75% of renewable energy by 2020, what is the necessary technology? This requires the 'new SCADA', and new agreements between customers and producers, so that the energy is used in the best way. And the most important issue is that this can be done without increasing the price of electricity.

The original Público story can be found here.

Related People:

Marija Ilic