Problem zone

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts
  • paulockenden
    Automated Home Legend
    • Apr 2015
    • 1719

    Problem zone

    My smallest room (downstairs loo) is having problems with temperature regulation.

    This is the temp as recorded by the HR92:

    IMG_0260.jpg

    And here's how open the valve is as it tries to compensate:

    IMG_0261.jpg

    I've tried resetting the zone, making the HR92 recalibrate, etc.

    Anyone got any other ideas?

    Like I said, the room is small but so is the radiator. I think the HR92 is just struggling to compute the amount of heat needed to keep a constant temperature.
  • paulockenden
    Automated Home Legend
    • Apr 2015
    • 1719

    #2
    Maybe I should play with the lock shield, and turn the flow down a bit. Or up a bit. Not sure....

    Comment

    • PaulB
      Automated Home Sr Member
      • Sep 2015
      • 60

      #3
      Originally posted by paulockenden View Post
      [ATTACH=CONFIG]863[/ATTACH]
      Out of interest, where are you getting the Percentage open variable? I can't find it anywhere in the JSON retrieved...

      Code:
              {
                  "alertSettings": {
                      "communicationFailureActive": true,
                      "communicationFailureMinutes": 15,
                      "communicationLostActive": false,
                      "communicationLostHours": 0,
                      "deviceID": 1303235,
                      "deviceLostActive": false,
                      "deviceLostHours": 0,
                      "faultConditionExistsActive": false,
                      "faultConditionExistsHours": 0,
                      "normalConditionsActive": true,
                      "tempHigherThan": 30.0,
                      "tempHigherThanActive": true,
                      "tempHigherThanMinutes": 0,
                      "tempLowerThan": 5.0,
                      "tempLowerThanActive": true,
                      "tempLowerThanMinutes": 0
                  },
                  "deviceID": 1303235,
                  "deviceType": 128,
                  "gatewayId": 1150889,
                  "holdUntilCapable": false,
                  "isAlive": true,
                  "isUpgrading": false,
                  "locationID": 742988,
                  "macID": "00D02D637898",
                  "name": "Living room",
                  "scheduleCapable": false,
                  "thermostat": {
                      "allowedModes": [
                          "Heat",
                          "Off"
                      ],
                      "changeableValues": {
                          "heatSetpoint": {
                              "nextTime": "2016-11-11T22:30:00",
                              "status": "Temporary",
                              "value": 21.0
                          },
                          "mode": "Off",
                          "vacationHoldDays": 0
                      },
                      "deadband": 0.0,
                      "indoorHumidity": 128.0,
                      "indoorHumidityStatus": "NotAvailable",
                      "indoorTemperature": 20.4,
                      "indoorTemperatureStatus": "Measured",
                      "isCommercial": false,
                      "maxCoolSetpoint": 90.0,
                      "maxHeatSetpoint": 22.5,
                      "minCoolSetpoint": 50.0,
                      "minHeatSetpoint": 15.0,
                      "outdoorHumidity": 128.0,
                      "outdoorHumidityStatus": "NotAvailable",
                      "outdoorTemperature": 128.0,
                      "outdoorTemperatureAvailable": false,
                      "outdoorTemperatureStatus": "NotAvailable",
                      "outdootHumidityAvailable": false,
                      "scheduleCapable": false,
                      "scheduleCoolSp": 0.0,
                      "scheduleHeatSp": 0.0,
                      "units": "Celsius",
                      "vacationHoldCancelable": false,
                      "vacationHoldChangeable": false
                  },
                  "thermostatModelType": "EMEA_ZONE",
                  "thermostatVersion": "00.00.00.00"
              },

      Comment

      • paulockenden
        Automated Home Legend
        • Apr 2015
        • 1719

        #4
        It's from the RF messages between the controller and HR92. I'm not using the API.

        Comment

        • g6ejd
          Automated Home Guru
          • Oct 2016
          • 153

          #5
          Originally posted by paulockenden View Post
          Maybe I should play with the lock shield, and turn the flow down a bit. Or up a bit. Not sure....
          There must be a point of maximum power (heat) transfer, open the LSV too far and does the hot water move through the radiator without time to transfer the heat, too little and the water flow is so slow that it dissipates too much heat but slowly, I think the physics or heat transfer has to be the same as power transfer and the best way to approach the problem is trial and error using 'limit's i.e. fully open then almost closed then iterate to a working solution is the way to approach the problem. Our so=called cloak room was like this until I fitted insulation min the space above it, now it only needs trickle of heat. Builders don't tend to insulate such spaces.

          Comment

          • paulockenden
            Automated Home Legend
            • Apr 2015
            • 1719

            #6
            Yes, the room loses heat. But as you can see from the actuator graph the HR92 doesn't realise this and keeps on ramping up and down, rather than just determining the flow needed and keeping there. It looks like a classic feedback loop to me.

            Comment

            • Little Tinker
              Automated Home Jr Member
              • Nov 2016
              • 38

              #7
              Originally posted by g6ejd View Post
              There must be a point of maximum power (heat) transfer, open the LSV too far and does the hot water move through the radiator without time to transfer the heat...
              That's kind of right. If the flow is very fast a litre of water enters and exits before it has much time to lose heat itself, but in turn the radiator is very hot all across it's surface so overall is radiating more heat than one that has slower moving flow making it hot at entry and cooler at exit. Crudely (and without crunching the numbers) imagine that on a slow flow that 1 litre passes by and loses 10 degrees temperature, and that with fast flow that 2 pass by and lose 6 degrees. Radiator is hotter end to end in 2nd case, but has lost 20% more heat to room.

              Comment

              • DBMandrake
                Automated Home Legend
                • Sep 2014
                • 2361

                #8
                Originally posted by paulockenden View Post
                Yes, the room loses heat. But as you can see from the actuator graph the HR92 doesn't realise this and keeps on ramping up and down, rather than just determining the flow needed and keeping there. It looks like a classic feedback loop to me.
                Yes it is a feedback loop, and in the case of a continuously oscillating room, its an unstable feedback loop. Warning, propeller head discussion of servo control systems theory and phase/gain margin in a feedback loop coming up.

                The HR92 uses a negative feedback loop to try to control the temperature - if the room is too cold it opens the valve a bit, if its too hot it closes it a bit, in a proportional way - simple right ? Unfortunately no. There's a whole field of maths around servo control systems and feedback loops and what it takes to make one stable.

                It turns out a negative feedback system like this can easily become unstable, when this happens it oscillates constantly at a certain frequency. To become unstable you need two things - 180 degrees of additional phase shift at some frequency (which turns your negative feedback into positive feedback) and a total gain in the feedback loop at that frequency where it reaches 180 degrees of greater than unity - eg amplification of the signal. When that happens the system becomes unstable and oscillates uncontrollably.

                Additional phase shift is caused by the response time lag of systems in the feedback loop. In the case of a radiator and HR92 you have a number of sources of time lag that contribute to additional phase.

                1) You have the thermal inertia of the radiator itself - it takes time to heat up and cool down. Heat up time is affected by whether the boiler is already up to temperature or not, (if it's not you have to add the time lag of the boiler heating up ) and it's generally a lot shorter for heating up than it is for cooling down - a radiator can be fully up to temperature in 2-3 minutes but can take 15-30 minutes to cool down again. This is very problematic from a servo control system point of view because if you make a mistake and open the valve too much or for too long, by the time you realise you made a mistake you will pay for it dearly with a temperature overshoot you can't control.

                2) You have the thermal inertia of the radiator heat transferring to the room. Even if the radiator was held at a constant temperature it takes time for heat to flow into the room and therefore time for the temperature to rise.

                3) The Honeywell system only samples the room temperature every 4 minutes. While this is OK for many rooms, in small rooms where the temperature can rise very quickly the reaction time of the sensors (due to the 4 minute sampling interval) is actually a bit too slow and in itself adds additional phase shift in the control loop that can lead to instability.

                So at some frequency the total phase shift will be 180 degrees making it possible for it to oscillate. The more the delays in the system the lower the frequency of oscillation will be.

                But to get instability and oscillation you also need greater than unity gain at the frequency where it would try to oscillate. A number of factors affect the gain of the feedback loop with an HR92 and radiator.

                1) The tuning of the gain constants in the PID control loop in the HR92 - we presume that it has self learning heuristics that can adjust the gain of the feedback loop in response to what it learns over time, however it will only be able to adjust its gain so far. It's worth noting that using "full stroke" effectively increases the gain of the HR92's feedback loop as it causes the motor to turn further for the same amount of temperature error, so full stroke = 1 would tend to make oscillations worse and that is my experience when I've tried it.

                2) The characteristics of your TRV valve - if the valve goes from off to full flow with say 2mm travel then the effective gain would be twice that if the pin had to travel 4mm to do the same. Some valves will be more proportional and some will tend to be more on/off and difficult to control precisely especially if they're sticky.

                3) Flow temperature - the higher your flow temperature the higher the total feedback loop gain because the same amount of movement of the motor controlling the valve will increase the heat going into the room much more at a higher flow temperature than at a lower flow temperature. A room that is unstable at a flow temperature of 75 might become stable and stop oscillating with a flow temperature of 60 for example as you'd be reducing the total feedback loop gain. I have noticed this with some of my rooms that are a bit marginal - set the flow temperature too high and it oscillates, drop it by 10 or even 5 degrees and it becomes stable.

                4) The size and type of the radiator relative to the room. A big radiator in a small room makes for a high feedback loop gain which can easily become unstable if there is also a significant response lag.

                For what its worth, three of my rooms do suffer from oscillations for one or both of the reasons I've described above. One is my kitchen:



                The reason this room overshoots and oscillates is because the radiator is boxed in between two kitchen cabinets and under a worktop, forming a small space that is partially cut off from the rest of the room with the HR92 stuck right in the corner of this box shaped area trying to measure the temperature. So every time the radiator heats up it causes a large quick increase in the temperature reading due to the boxing in of the radiator and temperature sensor.

                So in this case the instability is due to excessive gain - you get a really large increase in sensed temperature for a small opening of the valve. This causes more of an overshoot (high peak) and recovery than an oscillation.

                Here is the study:



                In this case there is a large radiator (about 1.6m x 600mm) that is at the far end of the house from the boiler which even with the lock-shield wide open takes a while to warm up. There is a significant delay in both heating and cooling with this large radiator and the result is too much phase shift causing instability, so you get a classic oscillation. Notice that it's symmetrical around the set point unlike the kitchen, which is a gain induced overshoot.

                Finally here is my bathroom:



                I suspect it has the same problem as your bathroom - a reasonably large radiator in a fairly small room means high gain to the feedback loop and a tendency to oscillate. The sampling rate of the HR92's (4 minutes) isn't fast enough to react to the quick temperature rise in a small room so some overshoot is inevitable.

                That's my take on the situation anyway, and none of my other zones have these problems. The Living room and Hallway in particular are spot on with almost perfectly flat temperature lines with no overshoot. Here is the living room:




                Does your bathroom still oscillate if you place a towel over the radiator to reduce the rate of temperature rise in the room ?
                Last edited by DBMandrake; 11 November 2016, 10:27 PM.

                Comment

                • Little Tinker
                  Automated Home Jr Member
                  • Nov 2016
                  • 38

                  #9
                  Originally posted by paulockenden View Post
                  My smallest room (downstairs loo) is having problems with temperature regulation.

                  I've tried resetting the zone, making the HR92 recalibrate, etc.

                  Anyone got any other ideas?

                  Like I said, the room is small but so is the radiator. I think the HR92 is just struggling to compute the amount of heat needed to keep a constant temperature.
                  I fitted my system last night, and our small upstairs and downstairs WC are the same. Downstairs is set to 21C, but showing an actual of 23.5C last time I looked (I have no logging yet). However, it doesn't feel that warm in there and when I bring the DTS92 we happen to have into there it is saying 21.5 rather than 23.5. But presumably the HR92 isn't self-correcting for that possible miscalibration .

                  I'm guessing it is because there isn't a lot of air flow in there, so the thermal mass of the water in the radiator after it turn off is slowly dissipating by hanging around the rad. If necessary I'm going to experiment with the lockshield, and maybe turning down boiler temp. Having only just got the system I'm maybe missing how it works, but if turning the rad down a couple of degrees still keeps it comfortable that would seem a valid way forwards as well .

                  Comment

                  • DBMandrake
                    Automated Home Legend
                    • Sep 2014
                    • 2361

                    #10
                    Originally posted by g6ejd View Post
                    There must be a point of maximum power (heat) transfer, open the LSV too far and does the hot water move through the radiator without time to transfer the heat, too little and the water flow is so slow that it dissipates too much heat but slowly, I think the physics or heat transfer has to be the same as power transfer and the best way to approach the problem is trial and error using 'limit's i.e. fully open then almost closed then iterate to a working solution is the way to approach the problem. Our so=called cloak room was like this until I fitted insulation min the space above it, now it only needs trickle of heat. Builders don't tend to insulate such spaces.
                    The issue of heat transfer versus flow rate and temperature drop across the radiator is an interesting one that I think is misunderstood in general.

                    There seems to be a general perception in the articles I've read that you have to have a measurable temperature drop across the radiator to show that heat is being radiated by the radiator - this is not the case at all.

                    You're partly right - like power transfer there will be an optimal flow rate that maximises transfer of heat from the water to the radiator and the room, but this is per volume of water flowing in a set amount of time.

                    Water has a finite specific heat capacity so if the flow is too slow you restrict the amount of heat being removed from the boiler, period - so the radiator won't get hot. If the flow is "too fast" you wont get any significant temperature drop across the radiator which means not much heat is being removed from a given volume of water but you have to remember that a greater flow of water means that there is more total heat transfer capacity available in the first place.

                    EG 20 litres flowing through the radiator in a minute can still transfer more heat than 5 litres flowing through per minute even though less heat per litre of water is being removed from the water at the higher flow rate. Make sense ?

                    What really matters is the heat transfer from the radiator to the room. At a low ("optimal" flow rate from the above) one side of the radiator will be at the full flow temperature, say 70 degrees, with a gradient across the radiator to the other side that might be at say 50 degrees due to the heat loss in the water as it travels across the radiator slowly. This gives an average surface temperature of 60 degrees and this will dictate the heat radiated into the room.

                    Now if you greatly increase the flow rate so that the temperature drop is only say 2 degrees across the radiator, now you have 70 degrees on one side and 68 degrees on the other side for an average of 69 degrees. You've just increased the average surface temperature by 9 degrees and more heat will be radiated into the room because of the smaller temperature drop across the radiator. Less heat will be removed from the water per litre of water flowing past but that is somewhat irrelevant as it is made up for by much more water flowing.

                    So from the point of view of maximising heat transfer from the boiler to the radiator and then the room, there is no such thing has a flow rate that is too high - the higher the better although you quickly reach a point of diminishing returns and run into other problems like cavitation and noise etc...

                    The real reason for trying to achieve a temperature drop across radiators is nothing to do with the heat transfer efficiency from water to radiator, its to drop the return temperature down sufficiently for a condensing boiler to be able to condense - that's where the efficiency increase of balancing the system properly comes from.

                    If you use a non-condensing boiler like me, there isn't really any point to trying to achieve a specific temperature drop across a radiator - as long as the radiators are hydraulically balanced so that no one radiator tries to steal all the flow and cause them to heat at different rates and the overall flow is sufficiently high that's all you need.
                    Last edited by DBMandrake; 11 November 2016, 11:15 PM.

                    Comment

                    Working...
                    X