Maximising Energy Savings Through Metered Data
(Note: this article makes references to diagrams which I cannot display here and modifications have been made to accommodate this
. The full article is available by email request)
To coin the clich, life, (like metering) is like a box of chocolates you never know what youre gonna get. For many organisations it is something of a no-brainer to have major and intermediate electrical loads metered either to a greater or lesser level of detail. For organisations with a more complex range of operations and processes, it can raise questions about the value of installing meters given the large capital investment sometimes required (I am here referring to the retrofitting of meters of course metering for new installations is covered quite comprehensively in the Building Regulations).
The returns on an investment on metering certainly vary on the users interaction with that metered data: at one end of the spectrum, having a plethora of meters sitting there taking measurements and with no one using the data, the metering installation will save you nothing. At the other end of the line, using every aspect of the data from one single meter can provide saving opportunities which could reduce the pay back period for the meter to one more preferable than the everyday investments a company makes.
This article looks at ways in which the return on the investment in metering can be maximised.
Why meter?
The typical response to the question posed Why should we meter energy anyway? has always been that you cannot manage what you cannot measure. However the question is frequently addressed with the notion in mind that you dont need metering in place to reduce energy consumption. Many people install energy efficiency into their own houses without even taking a glance at the electricity or gas meter. A much more accurate response to the Why meter? question would be that without it you cannot how energy-efficient plant
should be
could be
and actually is
This is very important as any resource available for improving energy efficiency in an organisation will be limited and time, effort and money should be focussed on those areas which will provide the largest opportunity for saving. Therefore knowing how good or bad your assets are in the first place will allow an energy conservation programme to be delivered in the most expedient way. It is pointless expending large amounts of energy, money and human resource in identifying and finding trifling energy savings.
What benefit?
As has been mentioned, the energy saving opportunities made available by metering are indirectly proportional to the interaction that the user has with the data.
I have identified four stages to the obtaining of useful data and savings made through the incorporation of an M&T system; there may be of course be more.
During the first stage of installing meters, naturally very few savings are identified albeit there may be some discoveries made during the installation that provide some savings through better knowledge of what assets there are.
The second stage is the very tempting one of looking at all the granular data in a very raw and unprocessed fashion. During this time some high-return quick-wins can be easily identified by someone who knows what they are looking for, including:
plant left on at night
plant left on at weekends/holidays
This can be a good place to stay for a while as there are often lots of opportunities to be discovered. It is also, however, a very easy place for the energy manager to get lost in and never be seen again as doing these spot identifications can be very labour intensive and often require some skill and knowledge to see them. Employing someone full time to go through a large site and keep track of these would be tempting but this goes against the compulsion of every energy manager: to devolve these tasks to the front-line workforce.
Benchmarking and exception reports start to rely on M&T software or even something as basic as a spreadsheet and this is where devolvement of overseeing the energy efficiency of running plant can begin. Furthermore, the nature of benchmarking and comparing similar assets begins to take us down the road of identifying optimum efficiency. Driving to achieve published best-practise targets can be interesting and challenging although it should be born in mind that some facilities may never achieve best practise performance due to the nature of their design or age.
Benchmarking can also be used to set up a competitive aspect within a multi-facility company where departments strive to improve their energy efficiency over others. Led properly and through the medium of a good behavioural change programme, this approach can deliver some substantial results as it engages many more people across the company. Benchmarking does rely on having knowledge of another variable quantity by which to normalise the energy data. This quantity is often known as the driving factor. Driving factors may be a constant for a certain asset (e.g. energy used per square meter of office space) or may in itself be a variable (e.g. energy used per degree day). Typical driving factors (expressed in their benchmark format) include:
By nature of business:
o Property energy/m2
o Manufacturing energy/item made
o Diverse commercial energy/ turnover
By nature of equipment:
o Air distribution energy/occupant/day
o Chilling energy/chilling degree day
o Heating energy/heating degree day
o Lighting energy/m2
o Small power energy/occupant or m2
o Personal transit:
Escalators energy/metre ascent
Moving walkways energy/metre travelled or per person
The list is fairly endless and there are alternative driving factors for most of the loads mentioned above. By having the same driving factors for similar equipment, comparison and exception reporting is far more meaningful than setting arbitrary targets.
Exception reporting can be incorporated to feed into a maintenance management system to dispatch technical support should certain pieces of equipment begin to run inefficiently. This is a fairly effective form of condition-based monitoring although shouldnt be relied upon for the availability of critical assets where more accurate and predictive techniques should be used such as acoustic, vibration or thermal monitoring.
It is at the stage of Benchmarking and Exception reporting that a true energy management strategy is called for. For the M&T initiative this may involve:
1. Identifying holes in the visibility of the data and installing more meters if necessary.
2. Grouping meters into asset types such as:
a. HVAC
b. Lighting
c. Small power
d. Processes etc
3. Analyse and compare similar asset types and their loads
4. Identify areas of inefficiency AND high efficiency (and ask the question Why?)
With comparison to gas consumption meters as well the range of quick wins and valuable questions include:
Why is the chilling fighting the heating?
Why is there so much lighting and small power left on at night?
Why is the building being air conditioned at nights and weekends?
Simply resolving the above three issues (and many buildings have them) can deliver between 10 20% savings.
The highest stage of analysis is the incorporation of driving factors (which may be used in benchmarking and exception reporting) into cumulative sum analysis (or CUSUM).
CUSUM Analysis
CUSUM is a type of dynamic benchmarking which seeks to resolve the problems inherent in normal benchmarking and exception reporting which include:
Ignoring small increases in consumption (creep) over a long period of time.
Distractions caused by temporary fluctuations.
Potential to miss important fluctuations (e.g. small step changes in consumption)
CUSUM does require having some data history of relatively stable operation of the asset under monitoring and is carried out in the following procedure.
1. Establish a record of driving factor against energy use
In the example we are looking at a piece of chilling equipment. We will avoid chilling degree days and simply look at peak external temperature against energy used for a sample of fifteen days.
2. Plot a scatter graph of driving factor versus energy used and establish a trendline.
The trend can be automatically generated with standard spreadsheet software. The relationship (R2 value) should be fairly good i.e. above say 0.5 and a reasonably steep line. If the relationship is poor this means either that the control of the equipment is very poor of you have selected the wrong driving factor!
Even at this stage, some fairly interesting information is available. Where the extension of the line crosses the y-axis, this quantity represents the Base or Standing Load and may have no correlation to the process being served. This amount is important as if it is very high then energy is simply being wasted or the process is not being controlled.
Before moving onto Cusum it is worth lingering a little more to look at this graphical analysis. By having a reference point of average performance, it is possible to identify times of good performance (those significantly below the line) and times of poor performance (those significantly above the line). As most spreadsheet software allows you to identify and examine individual plots, a bit of detective work looking into the past can inform the energy or facilities manager/technician what happened on those days. From this it is possible to eliminate reoccurrences of the bad days and replicate the circumstances for the good days. Ultimately we are seeking to do to things which will become visible on this graph:
1. Reduce standing load (drop the entire line)
2. Reduce process energy intensity (reduce the slope of the line)
It is unlikely, if not impossible, that either will be fully incorporated but a change in line position and slope in the right direction means better performance and better energy efficiency.
In one example from the authors experience, controls improvement works had, at one site, reduced the process intensity thus reducing the slope. The improvement works however had resulted in insulation lagging being removed and left off. The resultant trendline showed, as expected, a reduction in intensity but the intercept point had lifted signs of increased standing load.
3. Make predictions and record actual consumption
With a formula representing the trend line it is possible to make predictions of energy use once you know the values of the driving factor. By merely doing manual monitoring, this small step change may go unnoticed or be ignored once the system has settled into this new performance. The ongoing increase of energy used, or wasted, however mounts up over a period of time.
4. Calculate the variation between predicted and actual consumption
This is the linking step to full CUSUM analysis and is a simple one of merely subtracting predicted value of energy used from the actual value of energy used. The early part of the example shows small variations between the two with the positives more or less balancing out the negative values. This is expected and normal. Around Day 7 the difference between predicted and actual values becomes considerable even though it settles down to a reasonably level value (around Day 10-11). This is the first point at which the blip on the radar becomes clear.
5. Cumulatively sum the variations
Cumulatively summing the variations is a long way of saying, add up the differences. Doing this on a day by day basis shows very clearly when something has gone wrong. It also provides other indications such as:
Stable performance where stability occurs i.e. where the positive and negative fluctuations balance each other out
Problems stabilising when the problem occurred (a dramatic if not exponential rise in the CUSUM energy profile) and when it has become the norm (marked by a straightening albeit still upward-inclining line in the profile)
Effects of energy efficiency on stable operation (an effect of making the line horizontal and hopefully begin sloping down).
Typical Savings
The industry frequently suggests savings of between 10 and 20% achievable by installing and operating an M&T package including installation of meters and doing something with the data. This is of course based on the average result available from case studies. Even without sub metering it is possible to use such mathematical tools as CUSUM analysis at fiscal metering level to predict how much energy is being wasted and thereby what savings may be available once energy efficiency and some method of feedback is implemented. The figure of between 10 and 20% has been around for years and since then technology has evolved, occupants expectations have changed and utilities have become more expensive.
In recent projects potential savings as high as 40 and 50% have been identified. It becomes more difficult to evaluate exact savings from a metering and M&T project as the information that becomes available tends to pull together various initiatives, investments and efforts. One thing is for certain, the potential to install metering and see an unreasonable payback on investment is extremely low.
About The Author
I am the Executive Director of Sim Energy Ltd, an environmental consultancy based in Surrey and Oxfordshire in the UK.
I have a broad engineering background a specialism in energy efficiency after managing carbon and energy strategies for Heathrow Airport.
Our website address is cuturl('http://www.simenergy.co.uk') where you can find all contact details.
The author invites you to visit:
cuturl('http://www.simenergy.co.uk')
7 Must Have Online Data Recovery Tools