Wednesday, 9 September 2015

Heat Pump Performance Monitoring Examples

On 10th September I will be giving a brief presentation at the Ground Source Heat Pump Expo at the Ricoh Arena on the topic ‘Energy or Performance Monitoring’, so its timely to do a little blog here to elaborate on some of the examples I will be showing from some OpenEnergyMonitor dashboards I have been using

Note; The below are just examples for this blog, and don't necessarily show the whole story.

I have been working with OpenEnergyMonitor for some time, and now have various installs using OEM kit from Megni.co.uk
In brief, most of the systems I have installed use around 8 temperature sensors, CT power measurement with voltage sensing (real power including power factor), and/or pulse counting from standard electrical kWh meter.  I have used Grundfoss VFS flow sensors, but we are currently working on the direct interrogation of a Kamstrup heat meter, giving heat output

Data is sent via Ethernet to www.emoncms.org and displayed on dashboards (as samples below). These real-time graphs give a fantastic tool for the installer and the home owner. They show exactly what is happening now, and what has happened over the previous hour/week/month or year. The information can be used to improve the design of a system and also be used to fine tune the user settings.

Let’s start with a SIMPLE dashboard example

This type of dashboard can be accessed on any internet-connected computer using www.emoncms.org    e.g.  www.emoncms.org/example

The dashboard above shows a bar graph of daily energy input to the heat pump. This can be checked periodically for unusual values. It is showing high use on 15th March. By mousing-over the graph we can see that on this day, 24.2 kWh were used. The reason for this high use could be investigated.  The time period can easily be changed to anything you wish by zooming in or out.  Below this is the outside temperature. This might be interesting on its own right, but may be interesting if compared to energy used per day.   To the right are a few useful dials and figures -  cylinder temperature, room and outside temperature – things any home owner might like to know.    We can see that as from 6th April, the system is switched off.  
This type of simple dashboard is ideal for the home owner, but we can make as many dashboards we like, of varying complexity and detail.  These are very useful for installers and designers, and a far more in depth analysis can be made.

Let’s start with a good example of a GSHP connected to underfloor heating.

This is a 12kW (max output) inverter-drive GSHP operating for a 40 minute period here. The green area represents electrical input and the purple represents heat (direct reading from Kamstrup meter).  The ratio if these two areas gives the COP.  We can see the flow and return temperatures slowly ramping up to a final flow temperature of only 32°C.  Since this is September, the ground collector is exceptionally warm. This, along with the low flow temperature explains why the COP is currently almost 6.  Current conditions are ideal, but from tests earlier in the year, we expect to see average COPs for heating in excess of 4.
This graph is showing us a healthy flow-return dt of 6 degrees.  It is also showing how nicely the speed of the compressor drops in response to the rising flow temperature.


Below is another snip. This time from a fixed-speed GSHP. This one shows the source temperature too.

This shows a period of about 1/3rd of a day, and approx. 30min. run durations which is quite acceptable. The flow and return is nice and low, with average flow temperature around 30C.  The underfloor is a good design here, but he source is dipping below zero. This is not ideal, but a zoom-out of yearly temperatures and knowledge of total heat used would give better understanding.  In this case, since the underfloor is so good, it may be acceptable to have a slightly inferior ground source.

Next, a good example of a GSHP heating a domestic hot water cylinder.  The cylinder is copper with the heat exchanger coil in the bottom section of the cylinder. The heat pump is only 3.5kW and the coil is a nice large 3sq m.  


This example is showing the heat pump electrical input power as the shaded yellow area  (no heat meter fitted). The heating period starts at about 1.3kw input and finishes at around 1.7kw.  It also shows four temperatures;  cylinder top and bottom, and the flow and return temperatures.

This graph is showing the early evening heating period having been off by a time clock.
As we can see, the top of the cylinder is still at a useable 50°C before heating, but the bottom has dropped to 40°C.  The 24 minute heating period shown here starts by heating the bottom water from 40°C.  Indeed, this system has been set up carefully to ensure the system heats from a lower starting temperature.  The heat pump ‘sees’ flow and return temperatures of only 45/40°C at the start.  The 40°C cylinder bottom (not very hot) ‘pulls down’ the heat pump working temperatures, resulting in a high energy-efficiency.  By looking up the heat pumps performance data, we can estimate the average COP with reasonable accuracy; here it is about 3.5 at the start of the heating-up cycle.
As the cylinder warms, we can observe the point just before 18:00 where the bottom is becoming warmer than the top, and natural convection causes the top of the cylinder to rise with the bottom. After about 25 mins the whole cylinder has reached about 53°C.  At the end, the heat pump ‘sees’ temperatures of 55/52°C.  This is getting quite hot, and getting close to the limit of the heat pump’s comfort zone. The COP here may be about 2.8.  (taken from heat pump data).  We can then look at the period of time the heat pump has spent at different COPs, and estimate the COP for the whole DHW heating session. Its somewhere around 3.05.

If the system were enabled 24/7, and the sensor position not optimised,  the lower cylinder would not  drop so far, so the cylinder would  heat  more frequently from a higher starting point. The average working temperature would be much higher, so COP would be lower.  At worse, the COP could be not much better than 2.8.  Added to this, losses from the pipe run, and starting-up losses could result in worse performance.

We can therefore use the monitor to enable us to set the system to operate at a low average temperature, but for the cylinder top to remain at a useful temperature (e.g. say 47°C).
We can also see how nice and close the final cylinder temperature (53°C) is to the maximum flow temperature of the heat pump (55°C). This minimises the need for immersion heater (with COP of only 1).  In this case, the compact copper heat exchanger is exceptionally large compared to the heat pump size, and the coil is also only occupying the lower section of the cylinder.  This gives exceptionally good results, and allows us to heat some of the water  in a ‘batch’ from a colder starting point.

For the next example we have a complete contrast. This is a very inefficient system!

This one is a 14kW  ASHP.  The heat pump is fine, and function exceptionally well,  but the cylinder heat exchanger is debatably a little small for this big heat pump.  

Looking at the graph, heating starts when the middle of the cylinder is 48C.  the flow temperature s runs up to 60C within 15mins, at which point, the input power drops and the heat pump ‘tracks’ the 60C flow temperature.  After 30mins running, the cylinder is 55C.  The flow temperature here is considerable higher than our previous example. In part due to the heat pump NOT reducing its speed, and in part due to the smaller heat exchanger coil, but it has not done too bad.  
However, the period after 55°C is clearly grossly inefficient.    We can see that the compressor switches off frequently and spends the next 3.5 hrs! attempting to achieve 60°C.  The other thing to mention here is that the distance between the heat pump and cylinder is around 15m. What is actually happening is that most of the heat is simply being lost from the pipe run.  The energy consumed is shown by the yellow area of the power plot.  The final 5 degrees (to 60C) uses several times the power (area) of the first section from 1 to 2.   
The biggest problem here is poor use of the controls.Clearly, it would make a lot of sense to adjust the hot water setting to 55°C so that the heat pump stops.  .   
The final ‘floor heating’ period is just as terrible as the DHW period. Here, only 1 or 2 underfloor zones are open so the flow rate is far too low, as can be seen by the large temperature difference between the flow and return.  This is in excess of 10 degrees. Again, heat dissipated by the floor is far too small for this large heat pump.
This is a clear case of an over-sized heat pump connected to a cylinder and emitter system. A smaller unit would work far better.


Finally, just to top anyone up with a little heat pump theory, I am adding a graph that I used to illustrate heat pump efficiency v the output temperature.   If there is one thing to learn about heat pumps – LEARN THIS.

Here we have the characteristics of 2 sample heat pumps.  The vertical Y axis shows the efficiency, the COP. A 3kW immersion gives 3 kW of heat. It has a COP of 1. Heat pumps give out more heat than they consume because they extract heat from outside.  The X axis shows operating output temperatures ranging from tepid on the left to very hot on the right. 

I am showing 2 typical heat pumps. A typical (R407C refrigerant) unit can reach say 55°C, whilst a ‘high temperature’ (134A refrigerant)  unit may achieve 65°C.  Anyhow whatever type, we can see that to the left, where the water is lukewarm, the heat pump has an easy time, hence the COP is very high (1kw in for 4.5kW of heat). I liken this to driving a car up a slight incline. We should get good fuel economy here, maybe 50mpg .  However, it we heat up to 65°C, the temperature ‘lift’ is great, and this is a little like driving a car up a steep incline – we are in a low gear and the MPG is only 20!.    In the same way that you will NEVER get good fuel economy when driving up a very long steep hill, you will not get a good COP when heating to a high temperature.  That said, it should always be better than using an immersion heater.

So, knowledge of the performance of any heat pump should be understood, and data should be available for all models relating to output temperatures at specific ground source or air-source temperatures. 

If you have a high temperature heat pump, it doesn't mean you always have to operate it at a high temperature.  If you operate it at lower temperatures, the performance should be far better. It is however always a good idea to find out the working limits of you unit.

Some of you may have wondered about the flattening of the curve at very low temperatures.  I have drawn it that way because some heat pumps are not very good at the extremities (limits) of their performance. However, I am finding (and partly guessing) that most inverter heat pumps with electronic expansion valves work very well over a very wide range, so some heat pumps will easily exceed COPs of 5 in ideal conditions (usually late spring or autumn, when not much heating is needed) .Never forget mid-winter conditions - this is the time we need most heat, so the operational area to focus on.


So, it is with an understanding of the characteristics of heat pumps, performance monitoring can be used to great advantage. In general, we want heat pumps to spend as much time at lower output temperatures ( and high source temperatures) as possible.


Monday, 8 June 2015

Temperature sensing with OpenEnergyMonitor

Temperature sensors for monitoring heat pumps





I have been using OpenEnergyMonitor.org energy monitors with heat pumps for a few years now and thought I should briefly share my experience.  This covers the practical side of temperature sensors, and is based on my experience.
(If you just want advice about fitting temperatures sensors onto pipes, skip further down the page)

Introduction
The sensors used in OpenEnergyMonitor.org modules are DS18b20. These are Maxim 1-wire digital sensors.


They actually have 3 connections;   0v, 5v and signal.   Since each sensor has its own unique i.d. code, multiple sensors are simply connected into the same 3 terminals on the monitor device.
The great advantage of these digital sensors is that there are no errors due to cabling.  Phone extension cable can be good enough.  Thermistor sensors (variable resistance, as PT100) rely on a specific cable resistance and furthermore some sensors could ‘drift’ out of calibration.   A further advantage of digital sensors is that they do not in themselves generate heat. They can therefore be used in still air with good accuracy.
If several sensors are strapped together (with an elastic band) and tested in a thermos flask over a few hours, most seem to read within 0.2 degrees C of each other.  If you have say 6 sensors, you can chose those that agree most for the most important sensing (e.g. flow and return).   Anyhow, these sensors are perfectly accurate enough for this type of monitoring, and there is comfort in knowing that if you are seeing 35.2C (for example) on a graph, then that sensor tip must be at 35.2 +/-, with a small error margin of say 0.15 degrees.  This is because the digital signal is generated inside the sensor.

What to measure on a heat pump?

Typical temperature measurement could include the following:
Ground source
Water flow and water return from heat pump (hot side)
Ground collector glycol  inlet and outlet from heat pump (cold side) 
Outside (ambient) air temperature.  
Hot water cylinder temperature, and at least one room temperature.

Air source
Water flow and water return from heat pump (hot side)
Outside (ambient) air temperature and air-off (cooled air leaving heat pump) 
Hot water cylinder temperature, and at least one room temperature.

For either of the above there could be a buffer cylinder involved in the design, therefore the flow and return from buffer cylinder to emitter (e.g. underfloor or radiators) may need to be monitored.
There may also be mixing valves (sometimes unnecessary mixing valves) on underfloor manifolds. These may also need monitoring.

For more detailed analysis, the refrigerant internals of the heat pump can be monitored. These may include discharge, suction and liquid temperatures.   This is however a little intrusive, and could affect the warranty. It may be necessary to discuss this with the system installers.

Measuring air and liquids 

Air

Measuring internal air temperature is relatively straight forward, but sensors must be positioned away from any source of heat or radiation source.  Sunlight and close-by appliances and even lighting can greatly affect readings.
The general rule for outside air measurement is to keep the sensor well away from sunshine, or areas that get sunshine.  The sensor can also drop below the air temperature if it can ‘see’ a clear sky.  A small polystyrene roof is a good shield protection from positive or ‘negative’ radiation (i.e. when radiation from sensor is greater than its surroundings).  It is also necessary to keep the sensor dry, unless you purposefully wish to cover it with a wet ‘sock’ to record Wet Bulb Temperature.  There is a lot of information available about this relating to weather stations.


Liquids

Most heat pump monitoring involves the temperature measurement of water and liquids flowing in the pipes.
For very high accuracy results, as required by heat meter devices, an immersion pocket is required. (Heat Meters measure the difference between two sensors, so both must be very accurate)
The pocket usually involves a ‘tee’ fitting and a hollow pocket that is completely surrounded by the liquid. The sensor is inserted into the pocket.
  
An easier and cheaper method is to fix the sensor to the outside of a metal pipe. This method is used in all heat pumps for their on-board sensing.
Any sensor that is outside the liquid can be affected by the temperature of the surrounding air, but the error in the reading can be negligible if the sensor is mounted correctly.


There are various methods as follows-
1)  Sensor strapped to metal pipe or metal fitting
2) Sensor inserted in surface pocket (a pipe soldered to the pipe’s surface)


This shows 10mm pipes that have been soft-soldered to copper fittings. Due to the high conductivity of copper, the temperature of the inside of the small pipe is almost exactly the same as the temperature of the liquid inside the pipe. After insulating (lagging) this pipe, the accuracy of the sensor inserted inside the small pipe should be excellent.  Ideally the pocket internal diameter is close to the sensor diameter.  Conductive paste should be used before inserting the sensor
It is a good idea to fit several pockets like this in the right places during installation. Obviously the pipe must be empty.
Many heat pump manufacturers solder pipes on the outside of water or refrigerant pipes so that their sensor probes can easily be inserted to give accurate measurement


Strapping a sensor to a pipe (or a pipe fitting)

 Whilst not as good as surrounding the sensor with a conductive copper pocket, it is possible to simply strap the sensor to a pipe.   Strapping using a copper strap (as used in refrigeration) , or copper wire, will give better results.
The following shows one simple method.







Copper or brass conduct heat well.  The temperature difference between the inside of the pipe and the outside is tiny, so always fit sensors onto metals.  Plastic pipes are poor thermal conductors, so the effect of surrounding air can affect the sensor temperature. Furthermore, if the pipe is changing in temperature, the sensor may ‘lag’in time behind. i.e. it will respond very slowly to temperature changes.  This may or may not be a problem.

This is not the best position for a sensor since there is 3mm of plastic between the liquid inside and the probe, however, the copper wire strap is making the best of a bad job.  Heat will conduct along the copper. There is a lot of copper in contact with the plastic, and this will conduct and will transfer heat to/from the ‘surrounded’ probe.  If this is insulated well on the outside, it may give good-enough results for ground source comparisons.  Time response may be a little slow, which may not necessarily bad.   Fitting the probe onto a metal fitting is a far better bet.