RcWare Weather – Weather Forecast for Your Control System
Weather forecast for your control system
Savings, as seen from the point of view of building controls, mean proper maintenance of the control system and peripherials, optimum settings of system parameters, but also usage of suitable algorithms which contribute to maximum exploiting of the available energy. In other words, smart control strategy helps both saving energy and lower wearing of the system e.g. by less frequent switching periods of boilers, pumps, and chillers.
One of the frequently faced tasks is the effort to use the cheapest energy available, preferably the energy which is available for free in the environment. This energy is accummulated and then used at times when necessary. But how to estimate the amount of energy to be accumulated? The same problem appears in the heating circuits with long response times, such as Crittall heating. The response time is up to 12 hours and moreover, the heated rooms may be influenced by solar radiation, wind, and other factors.
When compiling the control strategy, a very helpful information may be that about weather conditions at the site for the next few hours or days. This may help to compensate the system response time to a certain extent. Some control systems (e.g. Sauter, Domat) already do have functions which read weather forecast data from the company servers in the Internet, but what can we do when a system is used which does not support this functionality, or when the software of old systems has to be modified during post-commissioning?
The solution is to use part of the open Domat control system, or rather its RcWare Weather service which reads the weather forecast data over the Internet, and brings the values into the existing system. In present time, the following values are available:
- U-component of wind (W-E direction, m/s)
- V- component of wind (N-S direction, m/s)
- Max. temperature 2 meters above ground (°C)
- Min. temperature 2 meters above ground (°C)
- Temperature 2 meters above ground (°C)
- Relative humidity (%)
- Precipitations (kg/m2)
- Downward long-wave radiation flux (infrared radiation, W/m2)
- Downward short-wave radiation flux (mostly visible light, W/m2)
- Total cloud cover (%)
- Pressure (Pa)
All values are predicted in a mathematical model for any given locality in Central and Eastern Europe, the system may be extended for other regions on demand. All variables belonging to one license are thus bound to a particular GPS location, defined when ordering the RcWare Weather service.
For each variable it is possible to define how many hours in advance the value should be predicted; usually, times from one to 72 hours are used. It is also possible to configure more variables reading one meteo value with different advance, e.g. outside temperature in one hour and outside temperature in six hours. The most used variable is outside temperature 2 meters above ground. At the example below there is optimizing of a heating curve preventing overheating: with expected rise of outside temperature the hot water temperature is set to a lower value in advance, to prevent unnecessary energy supply to the circuit.
The outside temperature at the site is measured with an outside temperature sensor, and the weather forecast is used only to predict relative temperature changes. This is because the locally measured temperature is mostly slightly different from the forecasted temperature – this difference is caused by sensor positioning and local conditions. The difference in predictions for t+3h and t+0h (i.e. temperature forecast for three hours later and for „now“) represents the expected temperature change, or trend. This change is conditionally added to the measured temperature: if the difference is positive, which means warming, the heating curve is fed with the corrected value. If the change is negative, or cooling is expected, only the measured temperature is taken into account. There is also communication check function in the system which secures that if the meteo data is not available (e.g. at internet connection dropout) the control system automatically works with classic heating curve dependent on the local measured outside temperature only. Usage of similar algorithm, slightly adapted to local conditions, lead to 16 % energy savings .
When a Domat IPLC510 process station or SoftPLC Runtime running on an industrial PC are used for controls, the meteo data are read from the driver and they can be used directly in the control algorithms. But even if there is another control system installed, it is possible to bring the values into it, and work in a familiar environment.
It is possible to communicate the predicted values in two ways:
- over analogue signals, and
- over a communication bus
The analogue signals, typically 0…10 V, are fine when the control system has no other way how to read the data. The signals are read from the output module(s): typically it is Domat M610, containing 8 outputs of 0…10 V, which can communicate up to 8 freely scalable variables. As a main processor, either a IPLC510 process station is used (Fig. 1), or a SoftPLC Runtime, running on any PC – e.g. a SCADA machine can be used (Fig. 2). The output module is then conected to the PC over a USB converter, which is more cost-effective than the IPLC-based solution.
Communication over a standard serial bus or network protocol means less hardware demands, because the output and input modules are not necessary. The existing control system, however, must support standards like OPC, Modbus TCP, Modbus RTU, or BACnet. The SoftPLC Runtime or IPLC510 process station then talk with the existing system over a RS485/RS232 serial line or Ethernet, see Fig. 3 for IPLC510, and Fig. 4 for IPLC510 or SoftPLC Runtime. No extra analogue inputs are required in the old system, which is useful if there are no reserve inputs, and new modules can not be added.
Even if an old control system is not capable of open communication, the battle may be won. The process station and Domat SoftPLC Runtime contains drivers for some frequently used control systems, so direct communication is possible e.g. over the Ethernet with Saia and Teco, and over a RS232 or RS485 serial line with GFR, Landis & Gyr PRU…, PRV… and RWP…, Sauter RSZ…, and Saia. This topology is drawn at Fig. 5. All the variants are free of extra costs, such as driver licences, etc.
At the time the IPLC510 or SoftPLC Runtime are configured, it is possible to perform auxiliary calculations, averaging, comparison etc., so the changes in the old system may be minimized. In extreme cases, when the old system uses a 0…10V active outside temperature sensor, this sensor may be replaced by the optimized predicted signal, so absolutely no changes in the old system are necessary.
Apart from heating circuits optimization, there are more aplications for using predicted meteo values: ice water storage control according to expected daily load, optimum AHU start / stop control (OSSC), watering, prediction of solar thermal and PV systems production etc. The process station is set up in factory so as to provide selected values scaled according to the customer specification, e.g. „output 1: outside temperature in the range of -30…50 °C“. The configuration may be changed later by simple editing of the SoftPLC application – the editing tool, SoftPLC IDE, is available for free. The accuracy of the forecast can be checked any time at http://rcware.eu/weather/prague/ with values for Prague – Czech Technical University, where also the measured temperature is recorded and compared with the forecast .
 Široký, Kubeček, Kudera: Prediktivní řízení vytápění budov – pilotní aplikace, TZB Haustechnik 3/2010