Using algorithms to operate heat pumps more efficiently and in a more grid-friendly manner

Heat pumps provide environmentally friendly heating and hot water. They are therefore a key technology for a climate-neutral building stock. However, an increasing number of these future-proof heating systems could place an excessive burden on the electricity distribution networks. The Centre for Solar Energy and Hydrogen Research Baden-Württemberg (ZSW) has now developed algorithms that reduce peak loads. The researchers tested the new method in Sweden, where heat pumps are already widespread and winters are particularly cold. The result: The algorithms help to operate heat pumps efficiently and in a way that serves the grid. The load on the transformers in the distribution grid decreased by ten percent.

Forecasts predict that the share of heat pumps in the German heating mix will rise sharply in the coming years. For low-voltage distribution networks in residential areas, this could become a burden without readjustment. Because: When it gets cold outside, all heat pumps deliver a high heating output at the same time – especially in the morning and at night. The demand for electricity in the distribution grid is increasing accordingly. Higher load peaks then occur in the grids and at the transformers that convert the voltage in the upstream medium-voltage grid to the voltage in the distribution grid. That could overload them.

The new method is one of many that the ZSW has developed for grid operators, manufacturers and users. The aim is to be able to use large consumers such as heat pumps and e-charging stations even in existing grids through intelligent operation, if possible without any noticeable restriction.

Heat pumps do not have to burden distribution grids

The researchers have therefore developed algorithms to reduce the simultaneity of heat pump loads in a grid area. “The challenge is to provide a warm house for everyone in the early morning and evening without all the heat pumps starting up at the same time – even on days when the outside temperature is minus ten degrees Celsius,” explains Dr. Jann Binder from ZSW. “To do this, we have developed a predictive heat pump operation that uses a forecast of heat demand.”

In the event of a foreseeable mains load, the heat pump switches on earlier and runs longer, but at a lower output. The process uses the heat capacity of the house as a storage medium and thus relieves the load on the grid. The researchers use this in a well-dosed manner so as not to increase the heat loss significantly and to keep the resulting temperature deviation from the setpoint within limits.

There were two approaches to choose from: a centralised approach, in which household heat pumps are incentivised to operate in a distributed manner by a central office via virtual energy pricing, and a decentralised approach, in which heat pumps simply respond to locally sensed temperature fluctuations and reductions in grid voltage, with no connection to a central office. The centralised approach achieves the required grid relief of ten percent with three percent less additional expenditure on heating energy than the decentralised approach, as it can avoid the need for “pre-heating” and the simultaneity of heat pump operation more precisely. However, it requires a large number of calculations to determine the individual schedules and thus more communication effort between all heat pumps and the control centre.

Ten percent less peak loads at the transformers

The result for the simpler decentralized approach: With the ten percent reduction of the transformer load at peak times, the spread of the indoor temperature changed only minimally; from 20 to 22 degrees Celsius to 19.2 to 22.2 degrees. If one additionally uses a forecast of the trend of the outside temperature, the lowest temperature is even limited to 19.4 degrees. If the same reduction in transformer load were to be achieved by linear reduction in heat pump output alone, the minimum indoor temperature would be 17 degrees, which is three degrees less, not just 0.6 degrees less.

When developing the decentralized approach, the ZSW paid attention to a simple design. “The algorithm does not need an external communication link for remote control of the heat pumps,” says Binder. “Locally measured line voltage is used as the source of information.” If the voltage falls below a limit value, this is an indication that the mains load is too high. As a result, the algorithm kicks in and modulates the heat pump output. Compared to a central control of heat pumps without complex bidirectional communication, a decentralized algorithm can use the ability of the house to store heat in an individual and well-dosed way. This reduces the resulting temperature drop compared to that which would occur if heat pumps were switched off centrally in the event of grid bottlenecks.

Test of the procedure in Sweden

In Sweden, the influence of heat pumps on the load on the electricity grid and their grid-serving operation can already be studied very well today. The Scandinavian country has a high carbon tax, so the use of heat pumps is already widespread.

The ZSW researchers chose the Ramsjö test site near Stockholm. Here, the houses are mainly heated with heat pumps. An ideal test area: In winter, the transformers were heavily loaded during particularly cold weather.

Why heat pumps are climate-friendly

Heat pumps are in vogue. Around 53 percent of all new buildings were equipped with the technology last year. The proportion is lower for the replacement of heating systems in existing buildings. However, their use will also increase significantly in renovated residential buildings in the future. The reason: on the way to climate neutrality, the heating sector must say goodbye to natural gas and oil.

Heat pumps help with this conversion. The devices obtain most of their energy from their direct environment – the air or the ground. The heat from the environment is renewable and available in practically unlimited quantities. To raise the temperature to the required level, heat pumps need electricity, which increasingly comes from wind energy and photovoltaic plants. This makes the technology more climate-friendly year after year. On average, three to five kilowatt hours of heat can be generated from one kilowatt hour of electricity per year, depending on the operating conditions and the technology used in the heat pumps.

The research project was part of the NEMoGrid project funded by the German Federal Ministry for Economic Affairs and Energy (BMWi) (funding code 0350016A). The term was just over three years and ended on 31 December 2020.