Skip to content
Joris Berkhout edited this page Nov 8, 2016 · 51 revisions

Table of Contents

Introduction

ETMoses relies on the use of time profiles to describe both the load that technologies exert on the network as well as the financial properties of the network over time. Profiles are currently provided with a time resolution of 15 minutes, but the calculation is versatile enough to deal with different time resolutions. The profiles heavily determine the meaningfulness of ETMoses results.

Types of profiles

Load Profiles

The load profiles determine the electricity load of base-loads and other technologies on a 15 minute basis. Various profiles are available by default.

Financial Profiles

The financial profiles determine the price of electricity on a 15 minute basis. Various profiles are available by default. These financial profiles can be used in a market model when the Price curve option under Tariff is selected.

  • There are several curves available to represent electricity prices that vary in time (e.g. Day-and-night tariff curves). Users can add their own profile for use in the market model. Currently, the available curves have been uploaded by users of the model and have not been researched or documented by Quintel.

Behaviour profiles

The behaviour profiles are binary profiles that can be applied to the hybrid heat pump. They specify when a hybrid heat pump must switch to using gas (i.e. when the profile equals 1). This is related to the outside temperature; when this temperature drops below 4 degrees, the COP of the electric part of the hybrid heat pump becomes too low. When the profile equals 0, a hybrid heat pump can run both on electricity or on gas, depending on the strategies applied and potential congestion of the network.

Examples of the profiles can be found on the website under Profiles -> Show all

Creating new profiles

Users can upload their own profiles for use in ETMoses; this works for load, financial and behaviour profiles. Each type of profile has its own option in the ETMoses profiles menu. Depending on the type, various attributes of the new profile have to be entered.

Load profiles

A profile can be uploaded to describe how the load of the technology varies in time. After uploading, the profile is normalised to unity (the area under the curve is 1) so the original units are irrelevant. Later, when creating your LES, the yearly demand or capacity of the technology is used to scale the profile.

The following attributes have to be supplied when creating a new load profile:

  • Key: a recognisable name for the profile
  • Name: optional display name for the profile
  • Default capacity: (OPTIONAL use with caution1) the default capacity (in kW) for the technology to which this profile will be assigned; this capacity will overwrite the default capacity of the technology itself
  • Default volume: (OPTIONAL use with caution1) the default volume (in kWh) for the technology to which this profile will be assigned; this volume will overwrite the default volume of the technology itself
  • Default demand: (OPTIONAL use with caution1) the default demand (in kWh) for the technology to which this profile will be assigned; this demand will overwrite the default demand of the technology itself
  • Permissions: select if profile should only be available to you or can be used by everyone
  • Included in concurrency: select if the profile can be included in the concurrency feature (see concurrency for more details)
  • Load profile category: select in which category the newly created profile should be listed
  • Curve: the actual load profile itself; the profile should have a 15 minute resolution and be specified for an entire year (i.e. contain 35040 data points)
  • Permitted technologies: select to which technologies the new profile can be assigned by the user

1 Some attributes have to be used with caution as they overwrite the corresponding attribute of the technology to which it is assigned and may therefore lead to unexpected results.

Financial profiles

A new financial profile requires three attributes:

  • Key: a recognisable name for the profile
  • Name: optional display name for the profile
  • Curve: the actual financial profile itself; the profile should have a 15 minute resolution and be specified for an entire year (i.e. contain 35040 data points)

Note that the unit of the profile should be EUR.

Behaviour profile

A new behaviour profile requires three attributes:

  • Key: a recognisable name for the profile
  • Name: optional display name for the profile
  • Curve: the actual behaviour profile itself; the profile should have a 15 minute resolution and be specified for an entire year (i.e. contain 35040 data points)

Note that these profile should be binary and can therefore only contain 0 and 1.

Default profiles

Quintel has researched and modeled various load profiles that are available to all users. In the sections below, the sources and modelling approach behind these profiles are described. The sections below follow the same structure as the list of all available profiles that can be found on the ETMoses website. Currently, the financial curves have been uploaded by users of the model and have not been not researched or documented by Quintel.

Load Profiles

Technology profiles

Base load

Within ETMoses, base load profiles are used to describe the load exerted on the network by all technologies within a household or building other than heat pumps or electric vehicles for which we use separate profiles. We distinguish two types of base load: inflexible and flexible. As the name suggests, the inflexible part cannot be moved in time or changed at all, because the technologies or appliances that cause this load (e.g. electric heaters and kitchen tools) need to be switched on at that time. The flexible profile contains the loads caused by washing machines, dish washers and dryers. The flexible load can be postponed by at most 3 hours (in the case of individual load profiles) or decreased to 0 (in the case of aggregated load profiles).

Individual
households

The individual load profiles for households are based on measured smart meter data that has been downloaded from the Liander website (under `Slimme meter -> Levering).

The smart meter load profiles have been processed in order to obtain an inflexible and flexible profile. The script that is used for this processing is available on GitHub.

buildings

No profiles have been researched or generated for individual buildings yet.

Aggregated
households

We use Energie Data Services Nederland (EDSN) profiles of type E1x and E2x as aggregated households load profiles. The profiles have been downloaded from the NEDU website. We use 2013 profiles.

buildings

We use Energie Data Services Nederland (EDSN) profiles of type E3x and E4x as aggregated buildings load profiles. The profiles have been downloaded from the NEDU website. We use 2013 profiles.

Electric vehicles

Because electric vehicles (EV) may be used for flexibility services (charging preferentially whenever there is local excess, discharging onto the grid etc.), a fixed load profile leaves insufficient freedom for the EV behaviour. Therefore, we have used created so-called availability profiles based on typical commuting behaviour (inspired by the report Laadstrategie elektrisch wegvervoer (Movares, 2013)) which leave considerable freedom in charging/discharging behaviour.

These profiles describe the minimum charge that needs to be present in the battery of the EV at any time and whether the EV is connected to a power outlet for (dis)charging. This information is used to determine when the EV needs to start charging as well as whether the battery in the EV can be used to store excess electricity produced by solar PV or to discharge when electricity is required locally, e.g. in the node that the EV is connected to.

We have generated 11 different EV availability profiles for 11 different driver types. We assume that each driver uses his car to commute to work on workdays and to go to other activities on both weekend days. The commute to work follows the same pattern every workday (i.e. the same departure and arrival times and distance travelled). In order to avoid the charging to start on exactly the same time every workday, we add a random time to the departure and arrival times in the interval [-30, 30] minutes. We do the same for weekend days. For weekend days we pick a random trip from a list of 14 'other' activities. Both the work and the data come from table D.1 in the PhD thesis of Remco Verzijlbergh (2013). Finally we take into account (public) holidays and sick leave by assuming that the EV stays at home for about 18% of the work days, selected randomly.

The profiles assume a charging power of 3.7 kW and battery volume of 100 kWh, although profiles can be created and uploaded for higher charging capacities (11 kW or 17 kW for example). Depending on the required charge the system works out when charging needs to start in order to fulfil future demand.

The availability profiles are used to determine the absolute minimum charge that needs to be present in the car battery to have enough "fuel" to run the next trip. The remaining battery volume could be used to store electricity (and use it later). From the required charge for the next trip and the charging rate, we calculate when the car has to start charging.

The availability is expressed as a percentage of the battery volume (i.e. 100 kWh).

As described above, for a given charging capacity, the main difference between profiles is in the distance of the commute. Hence, the names of the profiles are EV availability profile <S/M/L> commute <#>, where <S/M/L> is short, medium or long and <#> is a number. The table below specifies the exact distance of the commute for each profile.

Profile name Commute distance (km) Annual demand (kWh)
EV availability profile short commute 1 4.3 529.6
EV availability profile medium commute 1 24.8 1567.0
EV availability profile medium commute 2 31.0 1652.4
EV availability profile medium commute 3 35.2 1993.4
EV availability profile medium commute 4 40.1 2073.6
EV availability profile long commute 1 50.8 2447.4
EV availability profile long commute 2 58.8 2793.2
EV availability profile long commute 3 65.9 3330.9
EV availability profile long commute 4 77.0 3691.7
EV availability profile long commute 5 91.8 4300.5
EV availability profile long commute 6 109.2 5153.1

Solar PV

The solar PV load profiles have been downloaded from the SoDa website for the year 2013. To match the needs of ETMoses these profiles have been scaled such that their peak load is equal to 1.

Deprecated

This category contains older profiles that are needed for previously created LESs.

Miscellaneous

Contains profiles that do not fall in any of the other categories.

Space heating

These profiles can be associated with a space heating buffer and determine how this buffer will be 'emptied'. The technologies connected to this buffer need to make sure that the buffer is filled in time (see buffers for more details).

The default space heating profiles have been created by ECN and have been used by Ecofys in their study Waarde van slimme netten.

Hot water

These profiles can be associated with a hot water buffer and determine how this buffer will be 'emptied'. The technologies connected to this buffer need to make sure that the buffer is filled in time (see buffers for more details).

As we could not find any reliable, measured load profiles for domestic hot water (DHW) consumption, we have generated and uploaded a set of load profiles ourselves. To see which load profiles are available, see this issue on GitHub.

This script is used to generate typical time profiles for domestic hot water (DHW) demand. These profiles are generated for a period of one year and time resolution of 15 minutes. The script is based in great part on the study Realistic Domestic Hot-Water Profiles in Different Time Scales by Jordan (2001). Within this study the DHW demand is divided in four categories or types of events:

type A: short load (washing hands, washing food) type B: medium load (doing the dishes, hot water for cleaning) type C: bath type D: shower

Jordan (2001) defines probability curves for each type of events (see below for plots of the probability curves): these curves specify the probability of such an event to occur at every 15 minute time step. Combined with a typical volume per type of event and a typical number of occurrences per day, these probability curves are used to generate random DHW demand profiles. We have refined the method described in the study in three ways:

  1. we have adapted the volumes and occurrences to the Dutch situation as derived from other sources

    type volume per event daily occurrence per person
    A 1 5
    B 9 1
    C 120 0.143 1
    D 50 1

    1 once a week provided that the household has a bath, which we assume is true for 1 in every 7 households

  2. we have added the option to include events of type C (bath) in the DHW demand profile or not; these events represent relatively high hot water demands resulting in 'spiky' profiles which are not representative for most households as only 1 in 7 owns a bath

  3. we have made it possible to adapt the profiles to represent households with different numbers of people in it; some events happen more often when this number of people increases (e.g. showers), other events increase in volume (e.g. doing the dishes).

In addition we make the following two assumptions:

  1. we assume that all events occur entirely within a single timestep.
  2. we assume that all types of events require water at the same temperature; this means that the DHW demand in kWh is proportional to the DWH in liters. The script returns a normalised DWH demand profile that is scaled when used in ETMoses or ETModel.

The script returns randomly generated, normalised DHW demand profiles named DHW_demand_profile_<n>p_<b>_<i>.csv where <n> is the number of persons in the household, <b> is bath or no_bath and <i> is the number of the profile. In order to make sure that the same profile can be recreated, <i> is also used as a random seed while generating the profile. The script also returns a aggregated profile which takes into account the average number of people per household and the fact that only 1 in around 7 households has a bath.

Probability curves