11.07.2019 Views

Accurate Allocation of energy in production processes - the importance of using metamodels

Many companies struggle with correctly monitoring, allocating and distributing direct and indirect energy consumption over all internal and external customers, users or products. In this age of rising commodity costs and additional attention to sustainability, the advantages of adequately allocating energy are nonetheless substantial: 1. Real-time mapping of energy throughout process flows makes it easier to detect deviations in energy conversions during production processes, find the root causes and reduce losses. 2. Using exact and objective keys for allocating energy consumption speeds up periodic P&L and CSR reporting and makes for more accurate reports. 3. Continuous, real-time distribution of energy cost among users increases transparency and raises awareness with them. Although most companies already struggle with integrating correct data from metered equipment, this is only the first - albeit important - step. The crux is fully understanding the nature and efficiency of energy conversions processes; energy inflows and outflows to different customers; fixed and variable costs and corporate structures. By using metamodels - designed for and maintained by energy managers - all this information is added to the initial metered data and processed in such a way that any change from an energy point of view instantly leads to adjustments in the energy allocation. This white paper reviews emerging trends and challenges in energy allocation and examines the working and benefits of using specific metamodels at the level of energy management software.

Many companies struggle with correctly monitoring, allocating and distributing
direct and indirect energy consumption over all internal and external customers,
users or products. In this age of rising commodity costs and additional
attention to sustainability, the advantages of adequately allocating energy are
nonetheless substantial:
1. Real-time mapping of energy throughout process flows makes it easier
to detect deviations in energy conversions during production processes,
find the root causes and reduce losses.
2. Using exact and objective keys for allocating energy consumption
speeds up periodic P&L and CSR reporting and makes for more accurate
reports.
3. Continuous, real-time distribution of energy cost among users increases
transparency and raises awareness with them.
Although most companies already struggle with integrating correct data from
metered equipment, this is only the first - albeit important - step. The crux is
fully understanding the nature and efficiency of energy conversions processes;
energy inflows and outflows to different customers; fixed and variable costs
and corporate structures. By using metamodels - designed for and maintained
by energy managers - all this information is added to the initial metered data
and processed in such a way that any change from an energy point of view
instantly leads to adjustments in the energy allocation.
This white paper reviews emerging trends and challenges in energy allocation
and examines the working and benefits of using specific metamodels at the
level of energy management software.

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Condugo<br />

<strong>Accurate</strong> allocation <strong>of</strong><br />

<strong>energy</strong> <strong>in</strong> <strong>production</strong><br />

<strong>processes</strong><br />

The <strong>importance</strong> <strong>of</strong> us<strong>in</strong>g<br />

<strong>metamodels</strong>


Condugo<br />

<strong>Accurate</strong> allocation <strong>of</strong> <strong>energy</strong> <strong>in</strong><br />

<strong>production</strong> <strong>processes</strong><br />

The <strong>importance</strong> <strong>of</strong> us<strong>in</strong>g <strong>metamodels</strong><br />

Copyright © 2017 by Condugo<br />

All rights reserved. No part <strong>of</strong> this publication text may be uploaded or posted<br />

onl<strong>in</strong>e without <strong>the</strong> prior written permission <strong>of</strong> <strong>the</strong> publisher.<br />

For permission requests, write to <strong>the</strong> publisher, addressed “Attention:<br />

Permissions Request,” to <strong>in</strong>fo@condugo.com<br />

2


Many companies struggle with correctly monitor<strong>in</strong>g, allocat<strong>in</strong>g and distribut<strong>in</strong>g<br />

direct and <strong>in</strong>direct <strong>energy</strong> consumption over all <strong>in</strong>ternal and external customers,<br />

users or products. In this age <strong>of</strong> ris<strong>in</strong>g commodity costs and additional<br />

attention to susta<strong>in</strong>ability, <strong>the</strong> advantages <strong>of</strong> adequately allocat<strong>in</strong>g <strong>energy</strong> are<br />

none<strong>the</strong>less substantial:<br />

1. Real-time mapp<strong>in</strong>g <strong>of</strong> <strong>energy</strong> throughout process flows makes it easier<br />

to detect deviations <strong>in</strong> <strong>energy</strong> conversions dur<strong>in</strong>g <strong>production</strong> <strong>processes</strong>,<br />

f<strong>in</strong>d <strong>the</strong> root causes and reduce losses.<br />

2. Us<strong>in</strong>g exact and objective keys for allocat<strong>in</strong>g <strong>energy</strong> consumption<br />

speeds up periodic P&L and CSR report<strong>in</strong>g and makes for more accurate<br />

reports.<br />

3. Cont<strong>in</strong>uous, real-time distribution <strong>of</strong> <strong>energy</strong> cost among users <strong>in</strong>creases<br />

transparency and raises awareness with <strong>the</strong>m.<br />

Although most companies already struggle with <strong>in</strong>tegrat<strong>in</strong>g correct data from<br />

metered equipment, this is only <strong>the</strong> first - albeit important - step. The crux is<br />

fully understand<strong>in</strong>g <strong>the</strong> nature and efficiency <strong>of</strong> <strong>energy</strong> conversions <strong>processes</strong>;<br />

<strong>energy</strong> <strong>in</strong>flows and outflows to different customers; fixed and variable costs<br />

and corporate structures. By us<strong>in</strong>g <strong>metamodels</strong> - designed for and ma<strong>in</strong>ta<strong>in</strong>ed<br />

by <strong>energy</strong> managers - all this <strong>in</strong>formation is added to <strong>the</strong> <strong>in</strong>itial metered data<br />

and processed <strong>in</strong> such a way that any change from an <strong>energy</strong> po<strong>in</strong>t <strong>of</strong> view<br />

<strong>in</strong>stantly leads to adjustments <strong>in</strong> <strong>the</strong> <strong>energy</strong> allocation.<br />

This white paper reviews emerg<strong>in</strong>g trends and challenges <strong>in</strong> <strong>energy</strong> allocation<br />

and exam<strong>in</strong>es <strong>the</strong> work<strong>in</strong>g and benefits <strong>of</strong> us<strong>in</strong>g specific <strong>metamodels</strong> at <strong>the</strong><br />

level <strong>of</strong> <strong>energy</strong> management s<strong>of</strong>tware.<br />

3


Index<br />

Market drivers impact<strong>in</strong>g <strong>energy</strong> management and <strong>energy</strong> allocation 5<br />

The challenge <strong>of</strong> <strong>energy</strong> allocation mismanagement 5<br />

A brief history <strong>of</strong> <strong>energy</strong> management tools 7<br />

The solution: <strong>the</strong> metamodel as <strong>the</strong> core <strong>of</strong> <strong>energy</strong> management 8<br />

What exactly are <strong>metamodels</strong>? 9<br />

The core elements <strong>of</strong> <strong>metamodels</strong> 9<br />

The benefits <strong>of</strong> graph-based versioned models 12<br />

Hierarchical vs. graph-based <strong>metamodels</strong> 13<br />

A detailed graph-based versioned metamodel 17<br />

A model for accurate distribution keys 18<br />

Application: conversion efficiencies 19<br />

Application: cost allocation 19<br />

How to choose <strong>energy</strong> management s<strong>of</strong>tware 24<br />

Do I need a graph-based metamodel <strong>in</strong> my <strong>energy</strong> management s<strong>of</strong>tware? 25<br />

Do I need a versioned metamodel <strong>in</strong> my <strong>energy</strong> management s<strong>of</strong>tware? 25<br />

Condugo’s <strong>energy</strong> allocation module 26<br />

Co-creat<strong>in</strong>g <strong>the</strong> next generation <strong>of</strong> <strong>energy</strong> management s<strong>of</strong>tware 28<br />

4


Market drivers impact<strong>in</strong>g <strong>energy</strong> management<br />

and <strong>energy</strong> allocation<br />

We analysed 156 <strong>energy</strong> management applications for <strong>energy</strong> bookkeep<strong>in</strong>g.<br />

We found that <strong>the</strong> majority <strong>of</strong> <strong>the</strong>m do an excellent job <strong>in</strong> collect<strong>in</strong>g, validat<strong>in</strong>g<br />

and report<strong>in</strong>g straightforward meter<strong>in</strong>g data (ma<strong>in</strong>ly for electricity and gas),<br />

but that <strong>the</strong>y fail to provide mean<strong>in</strong>gful <strong>in</strong>sights as soon as <strong>energy</strong> conversions<br />

and chang<strong>in</strong>g company structures enter <strong>in</strong>to play.<br />

This is a cause for concern, as <strong>the</strong> general trend with <strong>in</strong>dustrial companies is to<br />

monitor all process flows - <strong>in</strong>clud<strong>in</strong>g conversions - and <strong>the</strong> associated <strong>energy</strong><br />

costs <strong>in</strong> real time and with <strong>in</strong>creas<strong>in</strong>g detail.<br />

In addition ‘<strong>energy</strong>’ is ga<strong>in</strong><strong>in</strong>g <strong>importance</strong> as a strategic issue <strong>in</strong> such companies<br />

and is no longer only <strong>of</strong> concern to <strong>the</strong> <strong>energy</strong> manager. Not only do several<br />

people throughout <strong>the</strong> organisation bear responsibility for a share <strong>of</strong> <strong>energy</strong><br />

consumption/costs, but <strong>the</strong> requirements on external report<strong>in</strong>g on <strong>energy</strong> and<br />

susta<strong>in</strong>ability have gone up as well.<br />

The challenge <strong>of</strong> <strong>energy</strong> allocation<br />

mismanagement<br />

The core challenge at <strong>the</strong> heart <strong>of</strong> <strong>energy</strong> allocation is to relate recorded <strong>energy</strong><br />

data to <strong>the</strong> specificities <strong>of</strong> <strong>the</strong> process flow and <strong>the</strong> organisation structure.<br />

Although this is a po<strong>in</strong>t <strong>of</strong> attention <strong>in</strong> any organisation, it is especially complex<br />

<strong>in</strong> <strong>in</strong>dustrial <strong>processes</strong>. This can be expla<strong>in</strong>ed by three factors: data issues,<br />

conversion <strong>of</strong> <strong>energy</strong> flows and corporate restructur<strong>in</strong>g.<br />

Data issues<br />

The logical approach to allocat<strong>in</strong>g <strong>energy</strong>, is to use a distribution key based<br />

on a comb<strong>in</strong>ation <strong>of</strong> ma<strong>in</strong> meter read<strong>in</strong>gs with submeter read<strong>in</strong>gs for each<br />

customer. This requires <strong>the</strong> <strong>in</strong>tegration <strong>of</strong> data from many meters, turn<strong>in</strong>g <strong>the</strong>m<br />

<strong>in</strong>to a consistent and permanent data stream <strong>of</strong> high quality to work with.<br />

Gett<strong>in</strong>g those data <strong>in</strong>, validat<strong>in</strong>g <strong>the</strong>m and correct<strong>in</strong>g any errors, is <strong>the</strong> first<br />

major challenge.<br />

5


Conversion <strong>of</strong> <strong>energy</strong> flows<br />

It becomes even more complicated when <strong>energy</strong> flows are be<strong>in</strong>g converted onsite,<br />

as is <strong>the</strong> case <strong>in</strong> many <strong>in</strong>dustrial companies. A common example is <strong>the</strong> one<br />

<strong>of</strong> CHPs. They take <strong>in</strong> gas and convert it <strong>in</strong>to heat and electricity, which could be<br />

used for produc<strong>in</strong>g hot water or power<strong>in</strong>g equipment. The consumption <strong>of</strong> an<br />

<strong>in</strong>ternal customer <strong>the</strong>refore consists <strong>of</strong> what <strong>the</strong>y consume directly plus <strong>the</strong>ir<br />

share <strong>of</strong> whatever went <strong>in</strong>to <strong>the</strong> conversions <strong>in</strong>to o<strong>the</strong>r <strong>energy</strong> types which<br />

<strong>the</strong>y consumed as well. Full understand<strong>in</strong>g <strong>of</strong> <strong>the</strong> nature and efficiency <strong>of</strong> such<br />

conversion <strong>processes</strong>, <strong>energy</strong> <strong>in</strong>flows, outflows to o<strong>the</strong>r customers, fixed and<br />

variable costs <strong>in</strong>volved, is crucial <strong>in</strong>formation that should be structured, added<br />

to <strong>the</strong> meter<strong>in</strong>g data and processed.<br />

This is not a trivial task and most companies stop short <strong>of</strong> this or rely on proxy<br />

measures to assign a share <strong>of</strong> <strong>energy</strong> used <strong>in</strong> conversions, to its users. Apart<br />

from be<strong>in</strong>g a source <strong>of</strong> <strong>in</strong>ternal discussions on P&L responsibilities, it frustrates<br />

any attempt to report on and track <strong>energy</strong> usage at <strong>the</strong> level <strong>of</strong> <strong>in</strong>dividual<br />

<strong>production</strong> batches.<br />

Correctly <strong>in</strong>corporat<strong>in</strong>g on-site <strong>energy</strong> conversions <strong>the</strong>refore is <strong>the</strong> second<br />

major challenge.<br />

Corporate restructur<strong>in</strong>g<br />

The third challenge is <strong>the</strong> consistency <strong>of</strong> <strong>the</strong> distribution key and overall <strong>energy</strong><br />

monitor<strong>in</strong>g when conditions change.<br />

Corporate restructur<strong>in</strong>g changes <strong>the</strong> entire fabric <strong>of</strong> departments and<br />

responsibilities. From an <strong>energy</strong> po<strong>in</strong>t <strong>of</strong> view this implies unbundl<strong>in</strong>g flows<br />

and reassign<strong>in</strong>g <strong>the</strong>m to different entities.<br />

Physical changes can be even more tricky: upgrad<strong>in</strong>g <strong>production</strong> <strong>in</strong>stallations,<br />

add<strong>in</strong>g or remov<strong>in</strong>g build<strong>in</strong>gs, reconvert<strong>in</strong>g entire sites - those are all reasons<br />

for add<strong>in</strong>g, remov<strong>in</strong>g or reconfigur<strong>in</strong>g <strong>in</strong>dividual <strong>energy</strong> meters.<br />

When such changes occur - which is more or less cont<strong>in</strong>uously <strong>in</strong> any larger<br />

organisation -, <strong>the</strong> new configuration <strong>of</strong> meters, <strong>processes</strong> and departments<br />

needs to be l<strong>in</strong>ked to <strong>the</strong> previous one to make report<strong>in</strong>g consistent and to<br />

cont<strong>in</strong>ue provid<strong>in</strong>g mean<strong>in</strong>g to <strong>the</strong> data.<br />

6


A brief history <strong>of</strong> <strong>energy</strong> management tools<br />

Craft<strong>in</strong>g appropriate distribution keys is <strong>of</strong>ten a tedious task that is undertaken<br />

once a year - with meticulous attention to details and hard work. Without <strong>the</strong><br />

appropriate s<strong>of</strong>tware tools, it is almost impossible to monitor <strong>the</strong>ir evolution<br />

permanently.<br />

Over <strong>the</strong> last ten years, <strong>the</strong> market for dedicated <strong>energy</strong> bookkeep<strong>in</strong>g s<strong>of</strong>tware<br />

has grown and <strong>the</strong>re are now a number <strong>of</strong> applications for collect<strong>in</strong>g, validat<strong>in</strong>g<br />

and report<strong>in</strong>g meter<strong>in</strong>g data. Most <strong>of</strong> <strong>the</strong> 156 <strong>energy</strong> management packages<br />

referred to above, <strong>in</strong>deed provide real-time monitor<strong>in</strong>g.<br />

When it comes to provid<strong>in</strong>g <strong>in</strong>sights and l<strong>in</strong>k<strong>in</strong>g meter<strong>in</strong>g data to <strong>the</strong><br />

organisation structure, however, only a very small number <strong>of</strong> those packages<br />

provide <strong>the</strong> option to do so. Deal<strong>in</strong>g with chang<strong>in</strong>g structures proved even<br />

more problematic, as none <strong>of</strong> those applications provided an option to take<br />

<strong>in</strong>to account chang<strong>in</strong>g structures without heavy custom development.<br />

Tak<strong>in</strong>g <strong>in</strong>to account <strong>energy</strong> conversions and allocation <strong>energy</strong> throughout such<br />

<strong>processes</strong>, turned out to be an even greater challenge that none <strong>of</strong> <strong>the</strong>m could<br />

address.<br />

The fundamental reason for <strong>the</strong>se shortcom<strong>in</strong>gs, is <strong>the</strong> lack <strong>of</strong> an adequate<br />

metamodel at <strong>the</strong> core <strong>of</strong> <strong>the</strong> current generation <strong>of</strong> <strong>energy</strong> management<br />

s<strong>of</strong>tware. Such a metamodel provides all contextual <strong>in</strong>formation that is needed<br />

to <strong>in</strong>terpret raw meter<strong>in</strong>g data. It describes <strong>the</strong> structure <strong>of</strong> <strong>energy</strong> flows<br />

(<strong>in</strong>flows, conversion, outflows); <strong>the</strong> structure <strong>of</strong> <strong>the</strong> company (<strong>the</strong> organisation<br />

chart); and <strong>the</strong> position <strong>of</strong> key people (those with responsibilities, as <strong>the</strong>y will<br />

need dedicated periodic reports) with<strong>in</strong> it.<br />

The <strong>metamodels</strong> <strong>in</strong> <strong>the</strong> s<strong>of</strong>tware - which were present <strong>in</strong> only a fraction <strong>of</strong><br />

<strong>the</strong> packages to beg<strong>in</strong> with - are hierarchical and static. A hierarchical model<br />

would give each asset a fixed position with<strong>in</strong> a set structure, where it belongs<br />

to one and only one higher category (e.g. a meter belongs to a build<strong>in</strong>g, which<br />

<strong>in</strong> turn could be part <strong>of</strong> a site). Energy conversions, however, break this mould<br />

as several commodities - and thus <strong>the</strong>ir meter read<strong>in</strong>gs - are comb<strong>in</strong>ed and<br />

recomb<strong>in</strong>ed <strong>in</strong> different configurations. As <strong>the</strong> models are also static <strong>the</strong>y are<br />

not aware <strong>of</strong> any changes that may have occurred, impair<strong>in</strong>g <strong>the</strong> ability to draw<br />

conclusions.<br />

All this po<strong>in</strong>ts at <strong>the</strong> dawn <strong>of</strong> a new generation <strong>of</strong> <strong>energy</strong> management s<strong>of</strong>tware,<br />

built around appropriate <strong>metamodels</strong> and able to allocate <strong>energy</strong> <strong>in</strong> chang<strong>in</strong>g,<br />

process-based sett<strong>in</strong>gs.<br />

7


The solution: <strong>the</strong><br />

metamodel as <strong>the</strong> core <strong>of</strong><br />

<strong>energy</strong> management<br />

8


What exactly are <strong>metamodels</strong>?<br />

‘To measure is to know’, is <strong>the</strong> adagio at <strong>the</strong> heart <strong>of</strong> <strong>energy</strong> management.<br />

Systematically collect<strong>in</strong>g <strong>the</strong> read<strong>in</strong>gs from a network <strong>of</strong> (sub)meters is <strong>in</strong>deed<br />

<strong>the</strong> start<strong>in</strong>g po<strong>in</strong>t. The boom <strong>in</strong> submeter<strong>in</strong>g <strong>of</strong> <strong>the</strong> last decade brought with<br />

it a proliferation <strong>of</strong> s<strong>of</strong>tware packages focused on collect<strong>in</strong>g, validat<strong>in</strong>g and<br />

visualis<strong>in</strong>g <strong>in</strong>formation. At <strong>the</strong> time <strong>of</strong> writ<strong>in</strong>g this document, Capterra lists<br />

156 applications under <strong>the</strong> head<strong>in</strong>g ‘<strong>energy</strong> management’. If we add generic<br />

enterprise data management s<strong>of</strong>tware be<strong>in</strong>g used for stor<strong>in</strong>g metered data <strong>the</strong><br />

tally goes up even more.<br />

Data management is important, but only a first step. Data needs to be <strong>in</strong>terpreted<br />

with<strong>in</strong> <strong>the</strong> specific structure <strong>of</strong> <strong>the</strong> company to really build understand<strong>in</strong>g <strong>of</strong><br />

<strong>the</strong> <strong>energy</strong> situation. Identify<strong>in</strong>g <strong>the</strong> real cost drivers or <strong>the</strong> most attractive<br />

improvement projects, follow<strong>in</strong>g up on performance - or assign<strong>in</strong>g <strong>energy</strong> to<br />

<strong>in</strong>ternal users - can only be done when <strong>the</strong>re is more <strong>in</strong>fo on <strong>the</strong> context. For<br />

each measured value, <strong>the</strong>re needs to be <strong>in</strong>formation on where and when it was<br />

recorded, what k<strong>in</strong>d <strong>of</strong> process/build<strong>in</strong>g <strong>the</strong> meter is associated with, what<br />

k<strong>in</strong>d <strong>of</strong> <strong>energy</strong> flow is be<strong>in</strong>g measured and how it is related to o<strong>the</strong>r meters<br />

and flows. Such ‘data on <strong>the</strong> data’ are what is called metadata. A structured<br />

overview <strong>of</strong> all relevant metadata and <strong>the</strong> relationships between <strong>the</strong>m, is what<br />

is called a meta-model.<br />

Most <strong>energy</strong> management s<strong>of</strong>tware is not built around <strong>metamodels</strong> and<br />

is <strong>the</strong>refore limited to record<strong>in</strong>g and visualis<strong>in</strong>g metered values, leav<strong>in</strong>g<br />

<strong>in</strong>terpretation entirely to <strong>the</strong> user/<strong>energy</strong> manager. A limited number <strong>of</strong><br />

programs -as mentioned before - do <strong>in</strong>corporate some k<strong>in</strong>d <strong>of</strong> model and<br />

allow for <strong>energy</strong> reports that are organisation-specific and <strong>in</strong>fused with<br />

mean<strong>in</strong>g. Typical examples <strong>in</strong>clude <strong>the</strong> monitor<strong>in</strong>g <strong>of</strong> <strong>the</strong> consumption <strong>of</strong> an<br />

entire build<strong>in</strong>g based on <strong>the</strong> read<strong>in</strong>gs <strong>of</strong> several submeters or <strong>the</strong> report<strong>in</strong>g on<br />

(variable) <strong>energy</strong> costs <strong>of</strong> an entire department.<br />

The core elements <strong>of</strong> <strong>metamodels</strong><br />

Complete <strong>metamodels</strong> for <strong>energy</strong> management conta<strong>in</strong> three basic categories<br />

<strong>of</strong> <strong>in</strong>formation.<br />

Structure <strong>of</strong> <strong>energy</strong> flows<br />

Every <strong>production</strong> process consists <strong>of</strong> a number <strong>of</strong> conversions, which toge<strong>the</strong>r<br />

transform flows <strong>of</strong> raw material <strong>in</strong>to a (semi-)f<strong>in</strong>ished product. This is reflected<br />

energetically as well. A number <strong>of</strong> <strong>energy</strong> flows are acquired or produced<br />

9


locally and <strong>the</strong>n used throughout <strong>the</strong> <strong>production</strong> process. In many cases <strong>the</strong>re<br />

will be conversions <strong>in</strong> between.<br />

Each submeter captures a part <strong>of</strong> this whole - and its place <strong>in</strong> it should be<br />

documented. This <strong>in</strong>volves both a specification <strong>of</strong> <strong>the</strong> type <strong>of</strong> <strong>energy</strong>/commodity<br />

it measures (electricity, gas, heat, compressed air,...) and <strong>the</strong> relationship to <strong>the</strong><br />

o<strong>the</strong>r meters. This way flows can be decomposed <strong>in</strong> <strong>the</strong>ir constituent parts and<br />

any losses/<strong>in</strong>efficiencies detected.<br />

Example graph based model with s<strong>in</strong>gle conversion<br />

Source node Distribution node Conversion node Distributionnode<br />

Natural gas grid<br />

Natural gas Hot water boilers Hot water<br />

connection<br />

S<strong>in</strong>k node<br />

Build<strong>in</strong>g product<br />

l<strong>in</strong>e 1<br />

S<strong>in</strong>k node<br />

Build<strong>in</strong>g product<br />

l<strong>in</strong>e 2<br />

Figure 1: Graph-based model example illustrat<strong>in</strong>g node types.<br />

Company structure<br />

Every meter ought to get a place <strong>in</strong> <strong>the</strong> overall company structure. They belong<br />

to departments, sites or build<strong>in</strong>gs. Groups <strong>of</strong> meters may toge<strong>the</strong>r capture <strong>the</strong><br />

<strong>energy</strong> situation <strong>of</strong> a part that is deemed relevant.<br />

Mapp<strong>in</strong>g <strong>the</strong> network <strong>of</strong> meters to <strong>the</strong> company structure is <strong>the</strong> bedrock for<br />

assign<strong>in</strong>g <strong>energy</strong> usage to cost centers throughout <strong>the</strong> company. Depend<strong>in</strong>g on<br />

<strong>the</strong> <strong>energy</strong> <strong>in</strong>tensity <strong>of</strong> <strong>the</strong> organisation, <strong>the</strong> impact on P&L can be considerable.<br />

A good metamodel captures this <strong>in</strong> a f<strong>in</strong>e gra<strong>in</strong>ed fashion, allow<strong>in</strong>g each<br />

<strong>in</strong>dividual meter to be clustered <strong>in</strong> several ways at <strong>the</strong> same time so as to<br />

reflect <strong>the</strong> complexity - and <strong>the</strong> peculiarities - <strong>of</strong> <strong>the</strong> company.<br />

Hierarchical model<br />

Company HQ<br />

Site Brussels<br />

Build<strong>in</strong>g 1<br />

L<strong>in</strong>e product A<br />

Build<strong>in</strong>g 2<br />

L<strong>in</strong>e product B<br />

L<strong>in</strong>e product C<br />

Site Antwerp<br />

Build<strong>in</strong>g 1<br />

L<strong>in</strong>e product D<br />

L<strong>in</strong>e product E<br />

Build<strong>in</strong>g 2<br />

L<strong>in</strong>e product F<br />

Site Ghent<br />

Build<strong>in</strong>g 1<br />

L<strong>in</strong>e product G<br />

L<strong>in</strong>e product H<br />

Build<strong>in</strong>g 2<br />

L<strong>in</strong>e product I<br />

Build<strong>in</strong>g 3<br />

L<strong>in</strong>e product J<br />

L<strong>in</strong>e pruduct K<br />

Figure 2: Hierarchical metamodel example for physical company structure<br />

10


Position <strong>of</strong> key people and responsibilities<br />

There are typically a number <strong>of</strong> people who bear some k<strong>in</strong>d <strong>of</strong> responsibility<br />

for <strong>the</strong> <strong>energy</strong> flows and costs <strong>of</strong> <strong>the</strong> company. Each <strong>of</strong> <strong>the</strong>m requires specific<br />

<strong>in</strong>formation and report<strong>in</strong>g, focused on <strong>the</strong>ir area <strong>of</strong> authority and tailored to<br />

<strong>the</strong>ir needs. The <strong>production</strong> manager will be more focused on permanent<br />

updates on <strong>the</strong> <strong>energy</strong> use and losses throughout <strong>the</strong> process; <strong>the</strong> purchas<strong>in</strong>g<br />

department’s first <strong>in</strong>terest is gett<strong>in</strong>g all <strong>in</strong>formation to periodically (re)<br />

negotiate <strong>the</strong> best contract price; account<strong>in</strong>g will be look<strong>in</strong>g at ways to verify<br />

whe<strong>the</strong>r monthly <strong>in</strong>voices match actual consumption; and <strong>the</strong> <strong>energy</strong> manager<br />

is on <strong>the</strong> lookout for new opportunities to reduce overall consumption. In<br />

organisations where <strong>energy</strong> has already become a strategic issue, <strong>the</strong> CEO will<br />

require periodic overviews <strong>of</strong> <strong>the</strong> evolution <strong>in</strong> total consumption, cost and CO2<br />

emissions - which will eventually show up <strong>in</strong> <strong>the</strong> annual reports.<br />

The people <strong>in</strong>volved and <strong>the</strong>ir precise positions <strong>in</strong> <strong>the</strong> company hierarchy,<br />

differ wildly. There is a clear trend, though, to have an <strong>in</strong>creas<strong>in</strong>g number <strong>of</strong><br />

employees need<strong>in</strong>g to pay some k<strong>in</strong>d <strong>of</strong> attention to <strong>energy</strong> consumption and<br />

costs; and to assign specific responsibility for <strong>energy</strong> improvement to a s<strong>in</strong>gle<br />

person - who may or may not be tak<strong>in</strong>g it up as a full-time job.<br />

A good metamodel <strong>in</strong>tegrates <strong>the</strong> <strong>in</strong>formation on those people. (Groups <strong>of</strong>)<br />

meters can be l<strong>in</strong>ked to one or more positions; access rights to each <strong>in</strong>dividual<br />

data po<strong>in</strong>t can be def<strong>in</strong>ed for each pr<strong>of</strong>ile; and <strong>the</strong> level <strong>of</strong> aggregation <strong>of</strong> <strong>the</strong><br />

<strong>in</strong>formation is <strong>in</strong>cluded.<br />

11


The benefits <strong>of</strong> graphbased<br />

versioned models<br />

12


Hierarchical vs. graph-based <strong>metamodels</strong><br />

The <strong>metamodels</strong> found <strong>in</strong> <strong>energy</strong> bookkeep<strong>in</strong>g packages are mostly hierarchical.<br />

This means <strong>the</strong>y categorize every data po<strong>in</strong>t - typically an <strong>energy</strong> meter read<strong>in</strong>g,<br />

but it could also be an <strong>in</strong>voice or a contract - as belong<strong>in</strong>g to a s<strong>in</strong>gle group or<br />

entity. This <strong>in</strong> turn is part <strong>of</strong> ano<strong>the</strong>r group or entity.<br />

The hierarchical metamodel is <strong>in</strong> very many ways a direct reflection <strong>of</strong> <strong>the</strong><br />

formal company structure. This is not unlogical:<br />

• Many packages are focused on build<strong>in</strong>gs and sometimes <strong>in</strong>dividual<br />

<strong>in</strong>stallations. Energy consumption is <strong>the</strong>refore broken down to that level.<br />

They are <strong>in</strong> one place and belong to one entity.<br />

• By reflect<strong>in</strong>g <strong>the</strong> formal hierarchy, <strong>the</strong> formal cha<strong>in</strong> <strong>of</strong> report<strong>in</strong>g is<br />

mirrored and data are likely to be available at <strong>the</strong> po<strong>in</strong>t where <strong>the</strong>y are<br />

needed. A department manager will need <strong>in</strong>formation on <strong>the</strong> overall<br />

<strong>energy</strong> consumption and contracts for <strong>the</strong>ir P&L; plus more f<strong>in</strong>e gra<strong>in</strong>ed<br />

output from underly<strong>in</strong>g submeters to p<strong>in</strong>po<strong>in</strong>t <strong>the</strong> ma<strong>in</strong> drivers <strong>of</strong> costs.<br />

The limitations <strong>of</strong> hierarchical models start show<strong>in</strong>g up when <strong>the</strong> focus shifts to<br />

<strong>the</strong> (<strong>production</strong>) <strong>processes</strong> ra<strong>the</strong>r than <strong>the</strong> build<strong>in</strong>gs or <strong>in</strong>stallations.<br />

Processes imply a sequence <strong>of</strong> <strong>in</strong>puts be<strong>in</strong>g comb<strong>in</strong>ed, converted and<br />

recomb<strong>in</strong>ed. All <strong>of</strong> those occur with a given efficiency, which can vary over<br />

time. The <strong>energy</strong> content or cost <strong>of</strong> a s<strong>in</strong>gle unit produced <strong>the</strong>refore is <strong>the</strong><br />

sum <strong>of</strong> (fractions <strong>of</strong>) <strong>the</strong> energies that went <strong>in</strong>to each <strong>of</strong> <strong>the</strong> preced<strong>in</strong>g steps.<br />

This can no longer be determ<strong>in</strong>ed by simply look<strong>in</strong>g at a s<strong>in</strong>gle meter output<br />

or a s<strong>in</strong>gle <strong>in</strong>voice; nor is it sufficient to aggregate a few meters. Depend<strong>in</strong>g<br />

on <strong>the</strong> process and <strong>the</strong> products <strong>in</strong>volved, datapo<strong>in</strong>ts will need to be grouped<br />

<strong>in</strong> specific ways and become part <strong>of</strong> more complex calculations. A s<strong>in</strong>gle<br />

datapo<strong>in</strong>t - aga<strong>in</strong>, most likely to be a meter - could <strong>the</strong>refore figure <strong>in</strong> different<br />

configurations, belong<strong>in</strong>g to multiple <strong>processes</strong> at once or be<strong>in</strong>g used several<br />

times over <strong>in</strong> a s<strong>in</strong>gle calculation.<br />

The solution is to move up one step <strong>in</strong> complexity to <strong>the</strong> <strong>in</strong>corporation <strong>of</strong><br />

graph-based models. They are far better <strong>in</strong> reflect<strong>in</strong>g <strong>the</strong> logic <strong>of</strong> <strong>processes</strong>:<br />

every entity is connected to a number <strong>of</strong> preced<strong>in</strong>g entities and a number <strong>of</strong><br />

entities succeed<strong>in</strong>g it. This is shown <strong>in</strong> figure 3 .<br />

13


Example graph based model with CHP and hot water conversions<br />

Source node<br />

Electricity grid<br />

connection<br />

Distribution node<br />

Electricity<br />

S<strong>in</strong>k node<br />

Build<strong>in</strong>g product<br />

l<strong>in</strong>e 1<br />

Source node<br />

W<strong>in</strong>d turb<strong>in</strong>e<br />

Conversion node<br />

Comb<strong>in</strong>ed heat<br />

power<br />

S<strong>in</strong>k node<br />

Build<strong>in</strong>g product<br />

l<strong>in</strong>e 2<br />

Source node<br />

Natural gas grid<br />

connection<br />

Distribution node<br />

Natural gas<br />

Conversion node<br />

Hot water boilers<br />

Conversion node<br />

Hot water<br />

Source node<br />

Water grid<br />

connection<br />

Distribution node<br />

Water<br />

S<strong>in</strong>k node<br />

Build<strong>in</strong>g product<br />

l<strong>in</strong>e 3<br />

Figure 3: Graph-based model example.<br />

As is immediately clear from <strong>the</strong> picture above, a graph-based model is not<br />

restricted to <strong>the</strong> one-on-one relationships between a datapo<strong>in</strong>t and <strong>the</strong> entity<br />

it belongs to. Interpret<strong>in</strong>g total <strong>energy</strong> consumption that went <strong>in</strong>to a batch on<br />

product l<strong>in</strong>e three, for example, is now possible as <strong>the</strong> model can trace back<br />

meter data on how much electricity (and from which sources) was used, which<br />

fraction <strong>of</strong> natural gas was used - both for heat<strong>in</strong>g water and for generat<strong>in</strong>g<br />

electricity - and how much water went <strong>in</strong>. The efficiency <strong>of</strong> <strong>the</strong> hot water<br />

distribution and <strong>the</strong> comb<strong>in</strong>ed heat-and-power (CHP) plant is readily available<br />

from <strong>the</strong> meter read<strong>in</strong>gs and can be taken <strong>in</strong>to account as well.<br />

Do note that hierarchical models can be seen as a less complex <strong>in</strong>stance <strong>of</strong><br />

graph-based ones. In a process where no <strong>energy</strong> sources are comb<strong>in</strong>ed or<br />

converted, every step gets only one <strong>in</strong>put from its predecessor. There would<br />

be only one arrow arriv<strong>in</strong>g at each node - which is <strong>the</strong> conceptual equivalent <strong>of</strong><br />

<strong>the</strong> datapo<strong>in</strong>t/node belong<strong>in</strong>g to a s<strong>in</strong>gle category.<br />

Static vs. versioned models<br />

A second perspective is to dist<strong>in</strong>guish between static models on <strong>the</strong> one hand<br />

and versioned models on <strong>the</strong> o<strong>the</strong>r.<br />

The few s<strong>of</strong>tware packages we found to have been built around <strong>metamodels</strong> <strong>in</strong><br />

<strong>the</strong> first place, were all based on static ones. The word ‘static’, however, does<br />

not necessarily mean <strong>the</strong> metamodel cannot change - <strong>in</strong>deed, one or two <strong>of</strong> <strong>the</strong><br />

packages did <strong>of</strong>fer (limited) options to adjust <strong>the</strong> configuration <strong>of</strong> datapo<strong>in</strong>ts/<br />

meters that are fed <strong>in</strong>to <strong>the</strong> program. They are still ‘static’ <strong>in</strong> our vocabulary,<br />

however, as <strong>the</strong>y lack <strong>the</strong> essential feature: <strong>the</strong>y do not track <strong>the</strong> changes that<br />

were at <strong>the</strong> basis <strong>of</strong> <strong>the</strong> change <strong>in</strong> <strong>the</strong> model. There is no way to uncover <strong>the</strong><br />

alterations to <strong>the</strong> structure <strong>of</strong> <strong>the</strong> <strong>energy</strong> flows, to <strong>the</strong> company structure or<br />

to <strong>the</strong> positions <strong>of</strong> key people. There are, <strong>in</strong> o<strong>the</strong>r words, no pathways that<br />

14


logically l<strong>in</strong>k <strong>the</strong> old and <strong>the</strong> new model to each o<strong>the</strong>r.<br />

The lack <strong>of</strong> pathways is a major impediment for <strong>energy</strong> report<strong>in</strong>g <strong>in</strong> complex<br />

and chang<strong>in</strong>g organisations.<br />

• Trend data are no longer mean<strong>in</strong>gful after a change. The <strong>in</strong>formation<br />

be<strong>in</strong>g reported - and <strong>the</strong> conclusions based on it - may or not be about<br />

<strong>the</strong> same entities or <strong>processes</strong> as before. The model itself will only<br />

allow for <strong>in</strong>terpretation based on <strong>the</strong> configuration based on <strong>the</strong> new<br />

situation, even when <strong>the</strong> bulk <strong>of</strong> <strong>the</strong> trendl<strong>in</strong>e is based on data be<strong>in</strong>g<br />

collected <strong>in</strong> <strong>the</strong> old situation. The issue is fur<strong>the</strong>r compounded when<br />

several, consecutive changes occur - each time wip<strong>in</strong>g out <strong>the</strong> mean<strong>in</strong>g<br />

that could have been given to older data read<strong>in</strong>gs.<br />

• The allocation <strong>of</strong> consumption or costs to entities, loses its value as<br />

a management tool. Whe<strong>the</strong>r <strong>the</strong> goal is to track performance, draft<br />

department P&Ls, do accurate cost account<strong>in</strong>g or <strong>in</strong>form CSR policy,<br />

<strong>the</strong> <strong>in</strong>formation ga<strong>in</strong>ed from a changed metamodel can no longer be<br />

put to use <strong>the</strong> same way as it may no longer cover <strong>the</strong> same underly<strong>in</strong>g<br />

elements.<br />

Versioned models <strong>of</strong>fer a solution. They are essentially a series <strong>of</strong> timestamped<br />

<strong>metamodels</strong>, each <strong>of</strong> which has been valid for a given period <strong>of</strong> time.<br />

Whenever a change occurs, <strong>the</strong> old version gets archived with <strong>the</strong> appropriate<br />

time stamp and <strong>the</strong> new one is generated. In addition, <strong>the</strong> l<strong>in</strong>ks between <strong>the</strong><br />

previous and <strong>the</strong> new model are made explicit. This means each new, changed<br />

or altered datapo<strong>in</strong>t must be highlighted as such; get its place <strong>in</strong> <strong>the</strong> overall<br />

structure and have its relationships with o<strong>the</strong>r datapo<strong>in</strong>ts def<strong>in</strong>ed. Examples<br />

<strong>in</strong>clude fitt<strong>in</strong>g an extra meter <strong>in</strong> a given process or add<strong>in</strong>g a new build<strong>in</strong>g to an<br />

exist<strong>in</strong>g department.<br />

Do note that a versioned model is a generalisation <strong>of</strong> a static one much <strong>the</strong><br />

same way a graph-based model is a generalisation <strong>of</strong> a hierarchical one. As<br />

long as <strong>the</strong>re are no changes to <strong>the</strong> underly<strong>in</strong>g <strong>energy</strong> or company structure,<br />

<strong>the</strong> version<strong>in</strong>g does not come <strong>in</strong>to play and <strong>the</strong> model is identical to a static<br />

one.<br />

15


The table below summarizes <strong>the</strong> ma<strong>in</strong> dist<strong>in</strong>ctions between <strong>metamodels</strong><br />

Hierarchical<br />

One category per<br />

datapo<strong>in</strong>t<br />

Graph-based<br />

Multiple precursors per<br />

datapo<strong>in</strong>t<br />

Static<br />

No awareness <strong>of</strong><br />

changes<br />

• Build<strong>in</strong>gs and stable<br />

tertiary sector<br />

environments<br />

• High-level <strong>in</strong>sight <strong>in</strong><br />

trends as long as no<br />

changes<br />

• Available <strong>in</strong> some<br />

s<strong>of</strong>tware<br />

• Stable <strong>in</strong>dustrial<br />

environments<br />

• Insight <strong>in</strong> trends lost<br />

as soon as changes<br />

• Only <strong>in</strong>directly<br />

available <strong>in</strong> a few<br />

applications<br />

Versioned<br />

Changes <strong>in</strong> model are<br />

tracked<br />

• Build<strong>in</strong>gs and<br />

chang<strong>in</strong>g tertiary<br />

sector environments<br />

• High-level <strong>in</strong>sight <strong>in</strong><br />

trends<br />

• Only available<br />

through heavy<br />

customization<br />

• Chang<strong>in</strong>g <strong>in</strong>dustrial<br />

environments<br />

• Full <strong>in</strong>sight <strong>in</strong> trends<br />

• Not available <strong>in</strong><br />

current <strong>energy</strong><br />

management apps<br />

16


A detailed graph-based<br />

versioned metamodel<br />

17


This section presents a concrete metamodel and uses it to illustrate how<br />

<strong>metamodels</strong> work, what <strong>in</strong>telligence <strong>the</strong>y embody and how much complexity<br />

<strong>the</strong>y capture.<br />

A model for accurate distribution keys<br />

The goal <strong>of</strong> this model is to provide an accurate distribution key for all<br />

commodity types present on <strong>the</strong> site <strong>of</strong> a company, especially when on-site<br />

conversions come <strong>in</strong>to play.<br />

We will refer to <strong>the</strong> example given below <strong>of</strong> <strong>the</strong> CHP and hot water conversions<br />

to expla<strong>in</strong> <strong>the</strong> model (See figure 4).<br />

Example graph based model with CHP and hot water conversions<br />

Source node<br />

Electricity grid<br />

connection<br />

Distribution node<br />

Electricity<br />

S<strong>in</strong>k node<br />

Build<strong>in</strong>g product<br />

l<strong>in</strong>e 1<br />

Source node<br />

W<strong>in</strong>d turb<strong>in</strong>e<br />

Conversion node<br />

Comb<strong>in</strong>ed heat<br />

power<br />

S<strong>in</strong>k node<br />

Build<strong>in</strong>g product<br />

l<strong>in</strong>e 2<br />

Source node<br />

Natural gas grid<br />

connection<br />

Distribution node<br />

Natural gas<br />

Conversion node<br />

Hot water boilers<br />

Conversion node<br />

Hot water<br />

Source node<br />

Water grid<br />

connection<br />

Distribution node<br />

Water<br />

S<strong>in</strong>k node<br />

Build<strong>in</strong>g product<br />

l<strong>in</strong>e 3<br />

Figure 4: Graph-based model example.<br />

The model requires three steps to achieve correct distribution <strong>of</strong> <strong>the</strong> commodities<br />

(<strong>the</strong> source nodes) to <strong>the</strong> customers (<strong>the</strong> s<strong>in</strong>k nodes).<br />

1. Partition<strong>in</strong>g source node flows over distribution nodes<br />

The first step distribute <strong>the</strong> consumption <strong>of</strong> each commodity<br />

(represented by <strong>the</strong> source nodes) over its direct customers<br />

(represented by <strong>the</strong> distribution nodes). This <strong>in</strong>itial partition is<br />

based on submeter read<strong>in</strong>gs for all customers <strong>of</strong> <strong>the</strong> particular<br />

commodity type (represented by <strong>the</strong> s<strong>in</strong>k and conversion nodes).<br />

In <strong>the</strong> graph model above <strong>the</strong> total commodity consumption<br />

equals <strong>the</strong> sum <strong>of</strong> all <strong>the</strong> <strong>in</strong>put flows <strong>of</strong> a distribution node.<br />

Total water consumption (a distribution node), for example, is<br />

what is provided through <strong>the</strong> water network (a source node).<br />

18


2. Allocat<strong>in</strong>g shares to s<strong>in</strong>ks and conversion nodes.<br />

As <strong>the</strong> goal is to get to <strong>the</strong> distribution over <strong>the</strong> f<strong>in</strong>al customers (<strong>the</strong> s<strong>in</strong>k<br />

nodes), <strong>the</strong> next step is to fur<strong>the</strong>r distribute <strong>the</strong> share <strong>of</strong> consumption<br />

allocated to <strong>the</strong> distribution and conversion nodes. This describes <strong>the</strong><br />

actual physical flow <strong>of</strong> <strong>the</strong> commodity types from <strong>the</strong> distribution node<br />

<strong>in</strong>to <strong>the</strong> o<strong>the</strong>r nodes (s<strong>in</strong>k and conversion nodes) . It is used to spread<br />

<strong>the</strong> total amount <strong>of</strong> commodity used over all <strong>the</strong> physical users <strong>of</strong> <strong>the</strong><br />

commodity type.<br />

3. Allocat<strong>in</strong>g <strong>in</strong>direct consumption from conversion nodes to s<strong>in</strong>ks<br />

The f<strong>in</strong>al step redistributes <strong>the</strong> flows that have gone <strong>in</strong>to conversion<br />

nodes. All users <strong>of</strong> <strong>the</strong> converted commodity, <strong>in</strong> o<strong>the</strong>r words, will get<br />

<strong>the</strong>ir share <strong>of</strong> <strong>the</strong> primary commodity <strong>the</strong>y <strong>in</strong>directly consume. This<br />

partition is crucial for transferr<strong>in</strong>g all costs - both direct and <strong>in</strong>direct -<br />

for all commodity types to <strong>the</strong> <strong>in</strong>ternal end customers <strong>of</strong> <strong>the</strong> company.<br />

The description above illustrates <strong>the</strong> power <strong>of</strong> graph models. The logic <strong>of</strong><br />

partition<strong>in</strong>g is already built <strong>in</strong>, so to speak. It def<strong>in</strong>es <strong>the</strong> <strong>energy</strong> type sources,<br />

conversions and s<strong>in</strong>ks for all <strong>the</strong> <strong>energy</strong> types provided by <strong>the</strong> company. Meters<br />

can be mapped to <strong>the</strong> flows <strong>of</strong> <strong>the</strong> graph model for all <strong>the</strong> periods relevant for<br />

<strong>the</strong> company - whe<strong>the</strong>r it is weeks or months - and <strong>the</strong>ir totals calculated over<br />

those periods.<br />

Application: conversion efficiencies<br />

The first application is straightforward. For each conversion node, <strong>the</strong> efficiency<br />

<strong>of</strong> <strong>the</strong> conversion is easily calculated us<strong>in</strong>g <strong>the</strong> <strong>in</strong>put and output flows. Every<br />

<strong>in</strong>put-output comb<strong>in</strong>ation gives rise to a particular efficiency.<br />

For example, a comb<strong>in</strong>ed heat power converter has a s<strong>in</strong>gle natural gas flow<br />

<strong>in</strong>put and two output flows, heat and electrical power.<br />

Application: cost allocation<br />

All costs for <strong>the</strong> acquired commodity types like, gas, water & electricity need to<br />

be allocated to <strong>the</strong> <strong>in</strong>ternal customers who use <strong>the</strong>m. In addition <strong>the</strong> costs for<br />

convert<strong>in</strong>g acquired commodity types to derived commodity types (e.g. hot<br />

water, result<strong>in</strong>g from <strong>the</strong> comb<strong>in</strong>ation <strong>of</strong> water and gas) need to be allocated<br />

to <strong>the</strong> <strong>in</strong>ternal customers that use <strong>the</strong>se derived commodity types.<br />

Direct costs are fairly easy to allocate, as <strong>the</strong>y can be l<strong>in</strong>ked to <strong>the</strong> meters which<br />

record <strong>the</strong> commodity type usage. The challenge is to correctly assign <strong>in</strong>direct<br />

19


commodity costs, like depreciation costs, operational costs and ma<strong>in</strong>tenance,<br />

to <strong>the</strong> <strong>in</strong>ternal customers. A graph-based model which <strong>in</strong>cludes commodity<br />

conversions will be very helpful <strong>in</strong> implement<strong>in</strong>g this application.<br />

All <strong>in</strong> all, three steps are needed to correctly allocate commodity costs to a<br />

company’s <strong>in</strong>ternal customers.<br />

Assign<strong>in</strong>g cost centers<br />

The first step <strong>in</strong> build<strong>in</strong>g a cost allocation application model is to assign cost<br />

centers to all nodes <strong>in</strong> <strong>the</strong> graph based model. Four types <strong>of</strong> nodes exist <strong>in</strong><br />

<strong>the</strong> graph model: source nodes, conversion nodes, distribution nodes and s<strong>in</strong>k<br />

nodes.<br />

Source nodes apply to acquired commodities on <strong>the</strong> one hand and locally<br />

generated ones on <strong>the</strong> o<strong>the</strong>r (e.g. electricity from w<strong>in</strong>d turb<strong>in</strong>es or solar plants).<br />

For acquired commodity types <strong>the</strong> monthly <strong>in</strong>voices will be assigned to <strong>the</strong>ir<br />

cost centers. For locally generated commodity types, <strong>the</strong> depreciation costs,<br />

operational costs and ma<strong>in</strong>tenance costs will be assigned.<br />

Conversion nodes are used for <strong>in</strong>dustrial <strong>in</strong>stallations which convert one or more<br />

commodity types <strong>in</strong>to o<strong>the</strong>rs. A typical example would be <strong>the</strong> conversion <strong>of</strong><br />

natural gas <strong>in</strong>to heat, which is distributed across <strong>the</strong> site via a pipel<strong>in</strong>e network<br />

<strong>of</strong> superheated pressurized water. The costs for <strong>the</strong> conversion <strong>in</strong>stallation like<br />

depreciation costs, operat<strong>in</strong>g costs and ma<strong>in</strong>tenance costs are assigned to<br />

<strong>the</strong>ir cost center.<br />

Distribution nodes are used to bundle all <strong>the</strong> sources <strong>of</strong> a s<strong>in</strong>gle commodity<br />

type and distribute <strong>the</strong> commodity among <strong>the</strong> o<strong>the</strong>r conversion, s<strong>in</strong>k or<br />

distribution nodes <strong>of</strong> <strong>the</strong> same type. Even when distribution nodes for <strong>the</strong><br />

same commodity type form a hierarchy, cost allocation is no problem as long<br />

as all <strong>the</strong> distribution nodes <strong>of</strong> <strong>the</strong> same type have <strong>the</strong> same cost center.<br />

Depreciation costs and ma<strong>in</strong>tenance costs for <strong>the</strong> distribution network on <strong>the</strong> site<br />

can be assignedto<strong>the</strong>ircostcenter. Typically<strong>the</strong>distributionnodewillalsodealwith<br />

losses<strong>in</strong> <strong>the</strong> distribution network which represent a cost. These losses will be<br />

distributed to <strong>the</strong> s<strong>in</strong>k nodes accord<strong>in</strong>g to <strong>the</strong>ir usage, ensur<strong>in</strong>g <strong>the</strong>se loss costs<br />

will be assigned to <strong>the</strong> <strong>in</strong>ternal customers automatically.<br />

For example a site converts natural gas and water <strong>in</strong>to steam. This steam is<br />

distributed over <strong>the</strong> site via a pipel<strong>in</strong>e network. This pipel<strong>in</strong>e network represents<br />

an <strong>in</strong>vestment which is depreciated on a regular <strong>in</strong>terval and will also require<br />

ma<strong>in</strong>tenance. Pipes corrode and need replac<strong>in</strong>g or valves break down and need<br />

replacement as well. T<strong>in</strong>y leaks <strong>in</strong> <strong>the</strong> pipel<strong>in</strong>e network cause pressure drops<br />

and represent losses. All <strong>the</strong>se costs are addressed by <strong>the</strong> distribution node.<br />

20


S<strong>in</strong>k nodes represent <strong>the</strong> <strong>in</strong>ternal customers <strong>in</strong> <strong>the</strong> company. These nodes have<br />

cost centers assigned to <strong>the</strong>m. The costs assigned to <strong>the</strong>se cost centers will<br />

- obviously - be <strong>the</strong> distributed costs from <strong>the</strong> o<strong>the</strong>r nodes based on <strong>the</strong> s<strong>in</strong>k<br />

node’s commodity consumption.<br />

Selection <strong>of</strong> <strong>the</strong> report<strong>in</strong>g period<br />

The second step is choos<strong>in</strong>g a report<strong>in</strong>g period. It is typically based on <strong>the</strong> firm’s<br />

overall f<strong>in</strong>ancial report<strong>in</strong>g schedule and not determ<strong>in</strong>ed by <strong>the</strong> requirements<br />

<strong>of</strong> <strong>energy</strong> management. Monthly or quarterly report<strong>in</strong>g periods are <strong>the</strong>refore<br />

common.<br />

The result would look like <strong>the</strong> picture below. All costs and losses for <strong>the</strong> selected<br />

period <strong>of</strong> time have been assigned to <strong>the</strong> nodes. They can now be used to<br />

allocate all direct and <strong>in</strong>direct costs to <strong>the</strong> s<strong>in</strong>k nodes.<br />

Example graph based model with CHP and hot water conversions<br />

Source node<br />

Electricity grid<br />

connection<br />

Usage: 100<br />

Gross<br />

prod: 150<br />

Loss: 0<br />

Distribution node<br />

Electricity<br />

Usage: 90<br />

Usage: 80<br />

S<strong>in</strong>k node<br />

Build<strong>in</strong>g product<br />

l<strong>in</strong>e 1<br />

Source node<br />

W<strong>in</strong>d turb<strong>in</strong>e<br />

Usage: 140<br />

Gross<br />

prod: 50<br />

Conversion node<br />

Comb<strong>in</strong>ed heat<br />

power<br />

Gross prod: 100<br />

Usage: 80<br />

Usage: 130<br />

S<strong>in</strong>k node<br />

Build<strong>in</strong>g product<br />

l<strong>in</strong>e 2<br />

Source node<br />

Natural gas grid<br />

connection<br />

Usage: 300<br />

Distribution node<br />

Natural gas<br />

Usage: 150<br />

Conversion node<br />

Hot water boilers<br />

Conversion node<br />

Hot water<br />

Loss: 10<br />

Usage: 100<br />

Gross prod: 150<br />

Loss: 10<br />

Usage: 80<br />

Source node<br />

Water grid<br />

connection<br />

Usage: 200<br />

Distribution node<br />

Water<br />

Loss: 50<br />

Usage: 80<br />

S<strong>in</strong>k node<br />

Build<strong>in</strong>g product<br />

l<strong>in</strong>e 3<br />

Loss: 40<br />

Loss: 10<br />

Figure 5: Graph-based model example with usages and gross <strong>production</strong>s..<br />

Distribution <strong>of</strong> commodity costs to <strong>in</strong>ternal<br />

customers<br />

For each report<strong>in</strong>g period <strong>the</strong> follow<strong>in</strong>g steps will be taken:<br />

1. Distribution <strong>of</strong> acquired commodity costs to <strong>in</strong>ternal customers,<br />

2. Distribution <strong>of</strong> generated commodity costs to <strong>in</strong>ternal customers,<br />

3. Distribution <strong>of</strong> converted commodity costs to <strong>in</strong>ternal customers,<br />

4. Translation <strong>of</strong> converted commodity usages to primary commodity<br />

usages,<br />

5. Distribution <strong>of</strong> commodity usages to <strong>in</strong>ternal customers,<br />

6. Total cost calculation per <strong>in</strong>ternal customer.<br />

21


Step 1: Distribution <strong>of</strong> acquired commodity costs to <strong>in</strong>ternal customers<br />

The acquired commodities like water, electricity and natural gas will have<br />

contracts and monthly <strong>in</strong>voices which specify both <strong>the</strong> costs as well as <strong>the</strong><br />

usage for each bill<strong>in</strong>g period. By divid<strong>in</strong>g <strong>the</strong> costs on <strong>the</strong> <strong>in</strong>voice by <strong>the</strong> usage<br />

for each commodity a commodity unit cost is calculated for each <strong>in</strong>voice period.<br />

For all commodity usages <strong>of</strong> acquired commodities <strong>the</strong> cost per period is<br />

calculated by multiply<strong>in</strong>g <strong>the</strong> usage by <strong>the</strong> unit cost.<br />

Step 2: Distribution <strong>of</strong> generated commodity costs to <strong>in</strong>ternal customers<br />

The generated commodities like electricity via w<strong>in</strong>d or solar, will have similar<br />

cost structures as <strong>the</strong> acquired commodity types. Depreciation costs, operat<strong>in</strong>g<br />

costs and ma<strong>in</strong>tenance costs which fall with<strong>in</strong> <strong>the</strong> current period are summed<br />

and divided by <strong>the</strong> gross <strong>production</strong> with<strong>in</strong> <strong>the</strong> same period. This yields an unit<br />

cost for <strong>the</strong> generated commodity types.<br />

For all commodity usages <strong>of</strong> generated commodities <strong>the</strong> cost per period is<br />

calculated by multiply<strong>in</strong>g <strong>the</strong> gross <strong>production</strong> by <strong>the</strong> unit cost.<br />

Step 3: Distribution <strong>of</strong> converted commodity costs to <strong>in</strong>ternal customers<br />

For <strong>the</strong> converted commodities, <strong>the</strong> cost structure is <strong>the</strong> same as for generated<br />

commodities: depreciation costs, operat<strong>in</strong>g costs and ma<strong>in</strong>tenance costs. The<br />

gross <strong>production</strong> is measured so <strong>the</strong> unit cost is as follows:<br />

For all commodity usages <strong>of</strong> converted commodities <strong>the</strong> cost per period is<br />

calculated by multiply<strong>in</strong>g <strong>the</strong> gross <strong>production</strong> by <strong>the</strong> unit cost.<br />

In addition to <strong>the</strong>se costs, <strong>the</strong> <strong>in</strong>put commodities for <strong>the</strong> conversion also br<strong>in</strong>g<br />

a cost. These will be calculated <strong>in</strong> <strong>the</strong> next step.<br />

Step 4: Translation <strong>of</strong> converted commodity usages to primary commodity<br />

usages<br />

Due to <strong>the</strong> <strong>importance</strong> <strong>of</strong> monitor<strong>in</strong>g <strong>the</strong> commodity conversion efficiencies,<br />

<strong>the</strong> <strong>in</strong>put commodity usages are measured directly and thus available for<br />

calculations.<br />

The costs associated with <strong>the</strong> <strong>in</strong>put commodity flows <strong>of</strong> <strong>the</strong> conversion nodes<br />

consist <strong>of</strong> <strong>the</strong> usage multiplied by <strong>the</strong> unit cost <strong>of</strong> <strong>the</strong> commodity.<br />

Depend<strong>in</strong>g on <strong>the</strong> source <strong>of</strong> <strong>the</strong> <strong>in</strong>put commodity flow, <strong>the</strong> unit cost is<br />

22


associated with an acquired commodity, a generated commodity or a converted<br />

commodity.<br />

Step 5: Distribution <strong>of</strong> commodity usages to <strong>in</strong>ternal customers<br />

Us<strong>in</strong>g <strong>the</strong> physical partition<strong>in</strong>g <strong>of</strong> <strong>the</strong> distribution nodes, both <strong>the</strong> usage and<br />

cost are<br />

distributed over <strong>the</strong> <strong>in</strong>ternal customers.<br />

Step 6: Total cost calculation per <strong>in</strong>ternal customer<br />

Internal customers, which are represented by s<strong>in</strong>k nodes <strong>in</strong> <strong>the</strong> graph model,<br />

use one or more commodities. The total cost for an <strong>in</strong>ternal customer is thus<br />

found by <strong>the</strong> some <strong>of</strong> <strong>the</strong> cost <strong>of</strong> <strong>in</strong>dividual commodity types.<br />

23


How to choose <strong>energy</strong><br />

management s<strong>of</strong>tware<br />

24


Not all companies have <strong>the</strong> same needs <strong>in</strong> terms <strong>energy</strong> management s<strong>of</strong>tware.<br />

The key question is how complex <strong>the</strong> relationship between measured <strong>energy</strong><br />

data, <strong>the</strong> process flow and <strong>the</strong> company structure, really is. This will determ<strong>in</strong>e<br />

how difficult it is to map process flows, detect deviations or allocate <strong>energy</strong><br />

consumption to specific users, products or batches.<br />

The guidel<strong>in</strong>es below will help you select <strong>the</strong> right k<strong>in</strong>d <strong>of</strong> s<strong>of</strong>tware.<br />

Do I need a graph-based metamodel <strong>in</strong> my<br />

<strong>energy</strong> management s<strong>of</strong>tware?<br />

When your company is only acquir<strong>in</strong>g commodities like natural gas, water and<br />

electricity and it is not convert<strong>in</strong>g those <strong>in</strong>to o<strong>the</strong>r commodities locally, a graphbased<br />

model is not strictly necessary. S<strong>of</strong>tware us<strong>in</strong>g traditional hierarchical<br />

<strong>metamodels</strong> can suffice. However, if you want to do cost allocation with <strong>the</strong><br />

<strong>energy</strong> management s<strong>of</strong>tware a graph based model can ease <strong>the</strong> setup <strong>of</strong> such<br />

an application.<br />

For example <strong>in</strong> <strong>the</strong> model <strong>in</strong> figure 3, <strong>the</strong>re are two conversion nodes: one CHP<br />

and one hot water. This company would def<strong>in</strong>itely pr<strong>of</strong>it from a graph based<br />

metamodel <strong>in</strong> <strong>the</strong>ir <strong>energy</strong> s<strong>of</strong>tware.<br />

Do I need a versioned metamodel <strong>in</strong> my<br />

<strong>energy</strong> management s<strong>of</strong>tware?<br />

Depend<strong>in</strong>g on <strong>the</strong> dynamics <strong>of</strong> your company, you may or may not need a<br />

versioned metamodel. Suppose <strong>the</strong> largest time frame on which you report is<br />

one year and <strong>the</strong> company structure - <strong>in</strong>clud<strong>in</strong>g sites, build<strong>in</strong>gs and product<br />

l<strong>in</strong>es - is stable with<strong>in</strong> <strong>the</strong> year, <strong>the</strong>n no versioned model is needed.<br />

If your company, however, launches two new products and builds two new<br />

product l<strong>in</strong>es dur<strong>in</strong>g <strong>the</strong> year, <strong>the</strong>n a versioned model with three versions may<br />

be needed to report accurately over <strong>the</strong> year. The first version describes <strong>the</strong><br />

start <strong>of</strong> year situation. The second version <strong>the</strong> new situation after <strong>the</strong> open<strong>in</strong>g<br />

<strong>of</strong> <strong>the</strong> first new product l<strong>in</strong>e. F<strong>in</strong>ally, <strong>the</strong> third version is needed after <strong>the</strong><br />

open<strong>in</strong>g <strong>of</strong> <strong>the</strong> second new product l<strong>in</strong>e. The year report<strong>in</strong>g period will need to<br />

be partitioned <strong>in</strong>to three partitions where each version is valid. After report<strong>in</strong>g<br />

on each partition, <strong>the</strong> f<strong>in</strong>al report will present <strong>the</strong> aggregation <strong>of</strong> <strong>the</strong> three<br />

partitions. When only <strong>the</strong> latest version is available <strong>in</strong> <strong>the</strong> s<strong>of</strong>tware <strong>in</strong> case <strong>of</strong> a<br />

non-versioned metamodel, <strong>the</strong>n no accurate report can be generated over <strong>the</strong><br />

year.<br />

25


Condugo’s <strong>energy</strong><br />

allocation module<br />

26


Condugo is a young company - founded <strong>in</strong> 2015 - <strong>of</strong>fer<strong>in</strong>g comprehensive<br />

<strong>energy</strong> management to <strong>in</strong>dustrial companies. At its core lies <strong>the</strong> modular<br />

Energy Hub platform, with tools and dashboards for monitor<strong>in</strong>g, analys<strong>in</strong>g and<br />

<strong>in</strong>terpret<strong>in</strong>g <strong>energy</strong> data.<br />

The platform’s Energy <strong>Allocation</strong> module has been designed around two specific<br />

models. The first one provides an accurate distribution key for all commodity<br />

types present on <strong>the</strong> site <strong>of</strong> a company, especially when on-site conversions<br />

come <strong>in</strong>to play. The second one allocates full <strong>energy</strong> costs down to <strong>the</strong> level <strong>of</strong><br />

<strong>in</strong>dividual <strong>production</strong> batches.<br />

Those models enable three types <strong>of</strong> analysis needed for advanced <strong>energy</strong><br />

management.<br />

• Conversion efficiencies and deviation analysis<br />

• Cost allocation from user down to batch level<br />

• Energy allocation <strong>in</strong> constantly chang<strong>in</strong>g environments<br />

The Energy <strong>Allocation</strong> module turns <strong>the</strong> power <strong>of</strong> those models and analyses<br />

<strong>in</strong>to a unique set <strong>of</strong> features :<br />

• Each metamodel has been designed for and can be ma<strong>in</strong>ta<strong>in</strong>ed by<br />

<strong>energy</strong> managers. Any change from an <strong>energy</strong> po<strong>in</strong>t <strong>of</strong> view with<strong>in</strong> <strong>the</strong><br />

organisation, is easily fed back <strong>in</strong>to <strong>the</strong> model and <strong>in</strong>stantly leads to<br />

adjustments <strong>in</strong> <strong>the</strong> <strong>energy</strong> allocation.<br />

• The module deals with any type <strong>of</strong> complexity. Smoothly handle any<br />

change related to (meter<strong>in</strong>g) data, history, conversion <strong>of</strong> <strong>energy</strong> flows,<br />

upgrad<strong>in</strong>g <strong>production</strong> <strong>in</strong>stallations, reconvert<strong>in</strong>g sites or corporate<br />

restructur<strong>in</strong>g.<br />

• Built-<strong>in</strong> <strong>in</strong>telligence. A s<strong>in</strong>gle datapo<strong>in</strong>t can figure <strong>in</strong> different<br />

configurations, belong to multiple <strong>processes</strong> or l<strong>in</strong>k to several key people<br />

at once. This guarantees <strong>the</strong> right people get <strong>the</strong> right <strong>in</strong>formation and<br />

report<strong>in</strong>g at <strong>the</strong> right time.<br />

27


Co-creat<strong>in</strong>g <strong>the</strong> next<br />

generation <strong>of</strong> <strong>energy</strong><br />

management s<strong>of</strong>tware<br />

28


Metamodels are start<strong>in</strong>g to show up <strong>in</strong> <strong>energy</strong> management s<strong>of</strong>tware, but only<br />

reluctantly. The need for packages built around more advanced (graph-based,<br />

versioned) models, however, is grow<strong>in</strong>g quickly. They are urgently needed as<br />

<strong>in</strong>dispensable tools for pr<strong>of</strong>essional and comprehensive <strong>energy</strong> management<br />

<strong>in</strong> complex <strong>in</strong>dustrial sett<strong>in</strong>gs, whe<strong>the</strong>r <strong>the</strong> goal is steer<strong>in</strong>g consumption,<br />

reduc<strong>in</strong>g costs or limit<strong>in</strong>g emissions.<br />

Condugo is build<strong>in</strong>g its Energy Hub platform to be this tool.<br />

After 1,5 years <strong>of</strong> <strong>in</strong>tensive development, grounded <strong>in</strong> cases <strong>of</strong> large and<br />

extremely complex <strong>production</strong> sites with extensive needs for report<strong>in</strong>g and<br />

cost allocation, <strong>the</strong> backbone <strong>of</strong> <strong>the</strong> s<strong>of</strong>tware is operational. In addition, first<br />

versions <strong>of</strong> its Monitor<strong>in</strong>g and Report<strong>in</strong>g modules have successfully been<br />

deployed with customers.<br />

The next step is to implement <strong>the</strong> Energy <strong>Allocation</strong> module on <strong>the</strong> platform,<br />

<strong>in</strong>corporat<strong>in</strong>g <strong>the</strong> models described <strong>in</strong> this white paper.<br />

As we strongly believe <strong>in</strong> co-creation, we have done extensive work with one <strong>of</strong><br />

our customers to def<strong>in</strong>e and design <strong>the</strong> module. We are now mov<strong>in</strong>g towards<br />

implementation and want to extend this co-creation effort to o<strong>the</strong>r <strong>in</strong>dustrial<br />

companies as well.<br />

That is why we call on o<strong>the</strong>r <strong>in</strong>dustrial companies to jo<strong>in</strong> us <strong>in</strong> this endeavour.<br />

Do not hesitate to get <strong>in</strong> touch to discuss <strong>the</strong> options and build <strong>the</strong> next<br />

generation <strong>of</strong> <strong>energy</strong> management s<strong>of</strong>tware, toge<strong>the</strong>r.<br />

29


<strong>in</strong>fo@condugo.com<br />

www.condugo.com

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!