Author Archives: Max

IBPSA NYC Chapter Mr.Comfy presentation posted for download

The presentation PDF of my NYC talk on Mr.Comfy is available for download or viewing on Scribd; it’s been a very enjoyable experience and I thank Pallavi Mantha and the IBPSA NYC chapter for this opportunity- and the entire Team at the EPFL who were such brilliant hosts when I presented it there the first time around.

Included in the presentation you find a few additional use examples of the daylight mapping overlay as well as previously unpublished student design mapping case studies; I will also post those separately bit by bit as the two main Mr.Comfy presentation are done for now (NYC and EPFL).


Mr.Comfy EPFL/IDEAS presentation, BSO14

Proud to report that Mr.Comfy will travel to the EPFL Lausanne with me, where I’ll hold a talk on spatial performance mapping in conceptual architectural design:

Here’s the poster (that is sure better than conference posters noone reads):

It will give me a chance to showcase some more of the spatial EnergyPlus and Daysim co-visualization class results as well as new features in the works (and already implemented since the last class ended), so I’m pretty excited about this.

Also, a paper on Mr.Comfy mapping methods has been accepted at Building Simulation and Optimization 2014, London; some of the content of that paper will already make it into the EPFL presentation. As always I’ll post both here once the show is over.

Work on new features is scheduled to begin again sometime in mid April; there’s some intricate stuff coming but I as of now have no idea when the next release will see the light of day- as ever, any feature requests, just drop me a line.



Mr.Comfy 0.2 released (daylight co-visualization)


Two different climate-based daylight mapping dot overlays on thermal data; frequency & average mapping shown with annual total zone cooling energy use

It has been a while since I had the chance to publish another version of the tool, and the feature I ended up putting in first again now somewhat surprised me, because originally I had other plans. However, experiences in the now wrapped-up spatial thermal mapping class made me want to experiment with other data co-mapping regimes before once again delving into the thermal nitty gritty. So version 0.2, as also requested by a few students, now offers basic climate-based daylight co-mapping features.

Co-mapping is really the word here, because the new component is not really designed to run on its own and without accompanying thermal input (though you can of course hide the thermal display if you want..). This is not really a coding rationale, but the idea behind the new tool is to do things differently from stand-alone daylight mapping aps, since you’re probably using Mr.Comfy to primarily look at thermal data in the first place and might want to see a quick daylight overlay when appropriate. In the 0.2 release, most of the daylight time inputs are therefore slaved to the thermal component inputs, with overrides exposed where appropriate.

As I’ve written here probably a few dozen times before, I usually use DIVA for daylight simulations and visualization, which is perfectly fine but for a few reasons made me wish for other ways of representing its (basically Daysim) simulation outputs. Firstly, it of course isn’t geared towards the simultaneous on-screen display of other (e.g. thermal) maps and thus obfuscates the Mr.Comfy metrics when you overlay them. Now that might not seem like a big issue, but design cognition works in subtle ways, hence the Mr.C daylight app uses a dithered “dot” representation style that always lets the thermal metrics shimmer through underneath. That gives quite a pleasant effect in simultaneously observing schedule-synched daylight and thermal results; also, since you can modify the dot size as you see fit, this kind of display works nicely for lower-resolution sensor node grids- it really makes sensors apparent as what they are, point-in-space readings.

Secondly, the most prominent DIVA metrics are formalized climate-based metrics flavors such as UDI, Daylight Autonomy etc. While those are well researched and used in practice, I always wanted to have a way to interactively overlay the DIVA metrics with “raw” sensor node data such as averages, custom frequency checks or even cumulative illuminance. Since the daylight tool exposes all the functionality of the thermal component, you can now easily achieve this.

Naturally, this is the first release of the daylight component, so a few fixes will follow. I guess including some scale manipulations to better get a handle on daylight data’s huge variability, and yes, speed improvements, are on the agenda. Speaking of speed.. it really is quite slow right now. That’s because daylight datasets can become very large very quickly (e.g. 20 million values is easy to achieve) and the code is not highly optimized (you have to start somewhere..). So for the time being I recommend you interrupt the processing of the daylight overlay when scrolling through the thermal representation.

For a detailed walk-through of the daylight functionality, please refer to the Climate-based daylight co-visualization User Guide page.

Lastly and on an entirely personal note, if you like this release or Mr.C in general, drop me a line. I’ve recently left my faculty position at the TU Berlin and am looking for new opportunities, so if you need a performance-minded person on your team,  please do get in touch. Thanks & happy co-mapping!

Spatial EnergyPlus visualization class result 01 & initial design cognition remarks

SophResult_lrAnnual heating peaks of Waratah Bay house, day/night schedule for different zones. Design, simulations and spatial metrics mapping: Sophie Barker

So the thermal spatial mapping class (already chronicled here, here and here) is more or less a wrap. Which begs the question whether we achieved what we set out to do- mainly to investigate if dynamic spatial visualization of E+ report variables does indeed improve architectural performance cognition during the design process.

The short answer to this is, maybe not surprisingly, a resounding “yes”, but it comes with a few caveats attached- the biggest one being that even if advanced visualizations are available, they exist in an ecosystem of related representations that also accompany the performance design process, e.g. your standard bar chart, daylight mapping etc., and those representations are mutually supportive. It still became rapidly apparent to me that students would more easily understand “classic” charts once they were pointed to “where and what to look for” in the maps- quite a nice result. A post-class survey additionally revealed that 100% of participants felt that they had learned new things about building performance simulation through dynamic spatial visualizations. Full results of all this will be published in an upcoming paper, which I will upload here as soon as I can.

SophResultB_lrSummer air temperature differences in Waratah Bay house, day/night schedule for different zones. Design, simulations and spatial metrics mapping: Sophie Barker

The first class result I’m sharing with you here maps seasonal performance for a house in Waratah Bay, Australia. It’s an interesting one because the house actually exists, and hence student Sophie Barker had quite a good feeling for when simulation results corresponded with reality (and when not). The main goal of the mapping exercise was to visualize annual air temperature comfort ranges and to determine the projected winter night heating energy consumption in bedroom zones. Summer natural ventilation vs. closed building states were also mapped, revealing that as already experienced in the existing structure, summer natural ventilation is enough to keep the main living spaces relatively comfortable. Download the full presentation PDF (13 MB).

What was neat about working with Mr.Comfy in this particular project is its ability to allow designers to think in temporal scenarios and ask specific questions such as “what’s the lowest hourly temperature in the bedroom zones in winter?” or “does it get uncomfortably hot in the living room during summer afternoons?”. That’s the sort of questions designers tend to ask and want answered in a visual fashion; hence I’m pretty happy with the results Sophie obtained in her project.

Another direct result from the class is that I’ve now also put in climate-based daylight mapping capabilities into the tool, since students often requested to be able to map those in a hybrid fashion, too, instead of relying on other software. So here’s the proof of concept blog post for that. Thanks for reading!

3d-printed energy models with Mr.Comfy & DIVA4Rhino

In the past classes hosted at the TU Berlin’s 3d lab, heavy use was made of rapid prototyping technology to create artifacts that show more than geometry only. Since architects think in models as well as drawings, it appeared to make sense to directly embed simulation data in physically manipulable ones, instead of just working with projective on-screen representations. I’ve had the great fortune to see many of the old daylight models featured on the DIVA frontpage, however the irradiation models shown here, created with DIVA4Grasshopper annual irradiation data, were never properly mentioned in any publication. You might notice some blue models hiding between the ones with a more familiar color range; those were 3d-printed within the same value scale as their sister models in different climate zones (we’ve got Iran, Sweden and Florida), so naturally the absolute irradiation they receive is less in some instances. Yes, the annual sum of solar gains is that much less in Northern Sweden than it is in Florida- one would, of course, assume so, yet to see it in color and physical 3d (practically better than IMAX) brings the point home somewhat more. Additional class details are given in the 2013 DIVA day presentation, describing the design development of two projects in Sweden and Iran; some musings on why such models are neat tools in a design process are further presented in my 2013 CAADRIA paper. Apart from being design aids, such models of course help communicating simulation results to others; especially when welcoming new students, curiosity about what they mean is instantaneous.

3d-printed urban irradiation models   3d-printed urban irradiation models; Sweden (Östersund), US (Hollywood, Florida, USA) and Iran (Yazd) sites*

Another artefact only briefly shown in the CAADRIA 2013 presentation (and not even in the paper) is somewhat of a departure from what had been done before; it’s a sectional model that shows facade irradiation in grayscale (the darker, the greater its intensity), combined with UDI100-2000 lux daylight metrics in the interior spaces. It was… somewhat difficult to build, quite unwieldy to handle and therefore an experiment not repeated, though as a demonstration object served its purpose well. It would be interesting to do something like it again, but with mapped thermal zone metrics instead of daylight- then one would really understand why certain spaces might suffer from their thermal plight (or perform well). But were mapped thermal metrics ever printed in a model? Sure, continue reading down below..

Hybrid, rapid-prototyped irradiation and daylight model3d-printed irradiation (facade) and daylight (UDI100-2000, zone-mapped) sectional model, Hashtgerd (Iran). Model, design & simulations: Piotr Jardzioch, Jakub Sobiczewski**

The last rapid-prototyped model fabricated in the now finished integrations project is a special one, since it for the first time shows the relative heating energy use of individual zones in a directly output physical model- before this, it had always only been the climate-based daylight distribution or irradiation metrics. Mr.Comfy made it possible to create analogous mappings for thermal data. The built design is a house in Southern Australia, modeled and simulated by Sophie Barker to discover which spaces might need the most thermal conditioning; I’ll post a more complete project description and a class summary in early 2014. What’s visible at first glance is that spaces with greater solar gains of course need less heating energy (it’s the Southern Hemisphere, so North is South, which I need to remind myself often enough) ; when random people handle such a model for the first time, it’s always astonishing to me how quickly they “get it”. A nice way to wrap up the TU project, for sure- off to new shores.

3d-printed energy modelRapid-prototyped model showing relative zone heating energy consumption;  Waratah Bay (Australia). Model & simulations: Sophie Barker

*thanks to all the past urban housing design class participants for producing such lovely irradiation models: Or Alexander Pearl, Dimitra Gkougkoudi, Piotr Jardzioch, Tereza Měřičková, Anastasia Vitusevych, Wolfgang Fischer, Rafael Kölmel, Caspar Kollmeyer, William George-Scott Sutcliffe

**special thanks to Jeffrey Tietze for having lots of patience with modeling support on this one


Mapping expectations / class impressions #3

The various class formats we experimented with through the years were in part always about how heuristic design thinking can be underpinned by simulation models as cognitive tools. It’s therefore a good exercise to finally be able to contrast performance expectations with actual mappings of simulation model outputs in a spatial format- centered around question/assumption complexes like “I assume it would be a rather lousy thermal situation in [insert building space here]. Let’s see whether that shows up in the model”. As it turns out, those assumptions usually do show up, connected with the discovery of other spatial performance pitfalls otherwise not easily diagnosed in large(ish) models. The two images accompanying this post are snaps from such an interactive question/answer session; the designer in question already suspected certain spaces of his design to be a tad problematic, something the other participants picked up on pretty quickly.


At any conference I’ve attended throughout the last two years, there has been a real glut of genetic/algorithmic performance optimization papers that claim to (semi)-automatically find various “optima” in a given design space, but few that actually take into account actual design thinking, which is still (and imho always will be) driven by the semi-heuristic knowledge base and experience of the involved designers. It always left me wondering whether once an “optimum” revealed itself, how robust would it really be, and to what degree would the people driving the machine actually understand why something is now “optimal”? As I wrote in an older paper for BS13, design thinking is (or at least can be..) holistic, expects the unexpected, and accounts for common sense performance modifiers that are not necessarily encoded in automated optimization procedures. Close the louvers, break performance; can your model close the louvers? Designer can imagine them closed, with the windows open, and disgruntled employees overriding the heating setpoints (maybe not the last bit, but we’ve all been there). What would happen then? The more performance domains interact, the trickier the whole situation. I’d argue that what needs to be the most robust is the mental model of performance interactions, something that can only be reinforced by properly re-representing domain baselines, and imagining a plethora of scenarios that might throw them into chaos. For once you have humans in the mix.. not all is linear. (go back in time to the previous class impressions post)


Mr.C on the E+ site! / Mr.C Whitepaper

Thanks to a friendly E+ team member who picked up my announce post on the bldg-sim list (thanks again!), Mr.Comfy has made its way onto the EnergyPlus tools directory:

I hope that quite a few more people will start using it- not aware of many just yet, apart from the class participants and a few engineering firms (but then no idea to what extent it’s being utilized). I could say now “it’s great and apparently has no bugs, so few people are in touch”, or… naw. It cannot be anything else. Of course.

Also added a new whitepaper publication to the bibliography. It’s the modified version of a contribution for an “IT in the AEC industry” competition, so I might as well get a mini paper out of it. German version only right now, but I guess a publication about Mr.C is coming up in ’14 anyways.

Thermal mapping class impressions #2

When we first started with the test class, I was wondering how the different mode of visualizing thermal results data would make an impact on performance understanding. We’d intensively worked with traditional line, bar, flood etc. charts/plots in previous classes, of course, and always observed that the extra level of abstraction- how to understand what’s going on in a building from a non-spatial chart, basically – made it ever so slightly more difficult for students and teachers alike to “get it”. Still, I always regarded traditional data representations as “hard and objective”, up to the point of beginning to second-guess whether the spatial mapping stuff I’ve worked on would be able to stand up on its own, and what other representations we might begin using alongside it.

I shouldn’t have worried, really. At the halfway point of this class, it’s become rapidly apparent to me that the design insights you gain through the spatial data representations -  in my humble opinion -  far outclass the avalanche of bar charts we’d relied on previously. That’s because you suddenly know exactly where stuff is happening, when it is happening and how it compares to other simultaneously occurring situations. If you’re in the business of tweaking and cajoling still pliable design states to make them perform better, that level of observational power seems to make a real difference. I hope to present class outcomes on this site when the show is over- at least one design will end up being published in a paper for BSO14- should peer review allow. (go back to the previous class impressions post | next class post)

thermal mapping class workshop, winter '13

thermal mapping class workshop, winter '13

Mr.Comfy 0.19 released

0.19 makes it possible to select time ranges that wrap around- yes, truly revolutionary coding at work here. You can now happily select all winter in one go, and map to your heart’s content. What made this the slightest bit tricky is that Python lists do not really wrap around when your last item access index falls off the cliff- hence Mr.Comfy computes the values for two slices if that’s the case, and adds them all up (or whatever it needs to do for the specific operation you request). Having several sets of slider inputs does not make this job easier, because the underlying code is unique for each set- you all better make plenty use of this functionality for slider animations, otherwise I’ll ditch it and give you one hourly input only. Kidding. Maybe.


AirFlowNetwork air change rate visualizations

I’ve never gotten around to posting about this until now, but when I first previewed the Mr.Comfy beta, my colleague Raoni Venancio – whose great work (Building Simulation 2011 paper, PDF) on how simulation influences design processes has been one of my frequent cites – remarked that it might be great to visualize air change rates with the tool as a sort of “quick and dirty pseudo CFD”.

The result is the image below, generated from data of an EnergyPlus AirFlowNetwork enabled natural ventilation run of the Cupertino sample building. The numbers are the average hourly air change rates for the whole year, in the subtropical climate of South Florida; second floor spaces exhibit much greater air change rates due to cross ventilation than the first floor offices. It’s not a super calibrated model, but we might get to the point of doing a better job visualizing air change rates for designs in the current test class, too- let’s see what comes out of it.