Author’s note: This is joint work with John Newhook, Department of Civil and Resource Engineering, Dalhousie University.
Over the past two years I have had the privilege to work on a number of different engineering initiatives related to the sport of curling. One of these is the development of an instrumented curling broom, a device that permits the measurement of force and stroke rate of a player, in real time, while brushing a curling stone. In this article, I’d like to present a brief overview of the body of work surrounding the study of brushing performance in curling, and briefly describe two research initiatives that are helping to increase our understanding of how we can improve brushing performance in athletes of all ages and abilities.
What makes a curling stone curl?
One would think that in a game played for the better part of 250 years with “in-turns” and “out-turns” we would have a solid grasp on the physics behind curling stones and a complete understanding of what makes a curling stone curl. The interesting thing about curling stones is that they nominally curl in the direction of rotation, whereas other objects (say a drinking glass on a kitchen counter) will “curl” on the counter-top in the opposite direction of the rotation. This phenomena has been studied, albeit incompletely, since the mid-1920’s and thus far the results remain controversial. Bradley  provides a brief summary of this research. An excellent overview can also be found in an article by Jon Minnaar of Scotland entitled “Why do curling stones curl?”, written in 2006, where the models of Shegelski and Denny are summarized. In 2013, Jessica Egan of the University of Utah published a brief survey of curling stone trajectory research; in a nutshell, research from the last 20 years or so encompasses a number of related but contradictory theories:
- Mark Shegelski of the University of Northern British Columbia, along with several different co-authors, has proposed a model [5,13-17] that is based on “wet” versus “dry” friction. In this model, the theory is that the stone’s leading edge forces a minute amount of water out of the ice, which lowers the coefficient of friction for the stone’s trailing edge and hence causes the stone to curl in its familiar direction.
- Mark Denny of BAE produced a model  based on left/right asymmetry of frictional force, based on the accumulation of ice (or other) debris on the surface of the stone’s running band through its travel, which he terms the “snowplow” model.
- Norikazu Maeno of Japan’s Hokkaido University proposes an “evaporation-abrasion” model [6,7] where the lower coefficient of friction for the rear portion of a curling stone’s running band is due to a combination of evaporation and ice debris generated by the front of the running band as the stone travels over the ice.
- A. Raymond Penner of Vancouver Island University proposed a slightly different model  than that of Shelgelski et al. in that he considered an “adhesive” effect of frictional force on a stone’s running band which results in a “pivoting” action of the stone as it rotates.
- Finally, recent work [10,11] by Harald Nyberg et al. at Sweden’s Uppsala University documents “micro-scratches” in the ice caused by the leading edge of a stone’s running band, and which in turn causes angular deflection of the trailing edge that causes the stone to “curl”.
The summary above was designed to be neither definitive nor exhaustive, but only to represent the uncertainty in our understanding of the physics of curling. That uncertainty leads to greater uncertainty when we consider what physics can explain about the brushing of a stone, something the Scots realized 150 years ago yet which we cannot – still – completely explain.
Development of instrumented curling brooms
It has been known for some time that the impact of brushing (or sweeping) in front of a curling stone lowers the coefficient of friction of the ice surface through the generation of heat [1,3]. Through the use of infrared cameras, a study undertaken for the Canadian Curling Association by Tom Jenkins at the University of Western Ontario in 2009 demonstrated this effect using Olympic-calibre athletes, and it was found that in the best case men could raise the surface temperature of the ice by between 0.75 and 1.75 degrees Centigrade, whereas women could raise the surface temperature between 0.50 and 0.75 degrees, primarily due to physiological disadvantages (strength, size, and body weight) compared to men. In coaching, however, it is important to provide more meaningful measures to athletes that are directly tied to their performance, so instrumented brooms capture metrics such as vertical force and stroke rate and use these as proxy variables, rather than the heat generated, in measuring athlete performance. Our specific interest is in determining which metrics to capture, the utility of each metric, and how one can use these metrics in the coaching of the sport.
To our knowledge, the development of the first instrumented curling broom (or “sweep ergometer”) was undertaken by Marmo, Buckingham et al. in Scotland [2,8-9] in 2005-6. The original prototypes captured information through a wired connection to a personal computer, rendering the device somewhat impractical but later versions transmitted the data wirelessly. This brush captured vertical force, stroke rate, brush head acceleration and velocity (used to compute the length of each stroke). Among other observations, the authors confirmed that the vertical force generated by an athlete is greatest when beginning the “push” portion of the stroke, and they used a mathematical model of the thermal dynamics of ice to illustrate the temperature impact during a stroke:
Marmo and his colleagues [8,9] were the first to document this phenomena in the scientific literature.
The second prototype comes from Japan . Hitoshi Yanagi and colleagues from the Kitami Institute of Technology developed an instrumented broom that captured horizontal and vertical force, stroke rate, and, perhaps most interestingly, broom handle angles and this data was compared to the data captured from a calibrated force plate:
A commercially available curling brush (Tapered Ultra-Light Carbon Fiber Brush, BalancePlus) was used to develop the device in the study. It consists of a head and a shaft with a joint combined to it. Eight strain gauges (N11-MA-10-1000-11, SHOWA) were attached to the shaft 200 mm away from the joint center to detect the loads in the direction of the shaft axis and in the direction perpendicular to it. Two angular velocity sensors (CRS07-02S, Silicon Sensing) were attached to the shaft to detect the shaft angle between the shaft and the ice surface and the roll angle about the shaft axis. The amplified signals from the strain gauges and the output signals from the angular velocity were digitized and sampled by the A/D converter at the rate of 200 Hz. The angles of the brush were calculated by using time integration of the angular velocities measured.
The forces on the brush were calculated as the forces applied to the ice surface vertically (FV brush) and horizontally (FH brush) based on the output from the strain gauges and the angles. The vertical (FV fp) and the horizontal ground reaction forces (FH fp) on the force plates (KYOWA) were also recorded at the rate of 200Hz during the sweeping simultaneously.
Unfortunately, the empirical studies that accompany the descriptions of both of these prototypes involve a very small number of subjects. One of the goals of our research includes the development of normative studies so that we can compare an athlete to norms for their gender, age group, skill level, or other criteria.
Commercial instrumented brooms
Quite recently, two commercial instrumented brooms have become available in Canada. The first is the PT-2 Smartbroom, developed in Kitchener, Ontario by Canada Curling Tools. The PT-2 features a re-chargeable, on-board integrated display with touch screen entry attached to a standard Balance Plus broom. The PT-2 captures both stoke rate (via an on-board accelerometer) and force, through two load sensors cemented into a customized Balance Plus EQ brush head.
The PT-2’s on board screen, shown in the photograph here on the left, can display not only summarized stroke rate and force statistics, but can also display a graph that plots these measures against time. The PT-2 saves time-averaged values of stroke rate and force every 0.5 seconds. The data can be downloaded off of the device via a USB connector and saved as CSV files that can be opened in spreadsheet software such as Microsoft Excel.
The second commercial instrumented curling broom, also Canadian, is from Sande Curling Tools of Winnipeg, Manitoba. Like the PT-2, the Sande broom captures stroke rate and force, although force is reported in PSI rather than in pounds as with the PT-2; additionally the Sande broom reports “amplitude”, or the length of the brush stroke, which we term displacement. Data from the broom is transmitted wirelessly to a laptop computer executing proprietary software that collects the data and summarizes it for reporting purposes.
Collaborative work on instrumented broom prototypes
At Conestoga, work on an instrumented broom prototype started in February 2013 when three students in the Computer Engineering Technology and Electronics Engineering Technician diploma programs took on the challenge of fabricating a working prototype broom to capture stroke rate as their project in a 4-day Engineering contest.
The Conestoga broom, shown in action at left, transmits its data via Bluetooth to an Android smart phone, using a display program developed by Marc Bernard, one of the coaches at Wilfrid Laurier University. Emerging from that contest experience with a win, students Aaron Schryver and Brandon Davies subsequently represented Conestoga at the November 2013 Polytechnics Canada conference, held at the Southern Alberta Institute of Technology in Calgary, where the prototype broom was featured in a Global News summary of the two day event.
Since December 2013 I have been delighted to collaborate with John Newhook of Dalhousie University regarding both the design of instrumented curling brooms and their application in a coaching context. The many hats John wears include varsity curling coach at Dalhousie, and Professor of Civil and Resource Engineering. John and his research team at Dalhousie have also had a prototype instrumented broom under development, called the CurlSmart broom, now in its sixth iteration.
As with the commercial products, the CurlSmart device can measure force and stroke rate, amongst other metrics. Data from the device is transmitted wirelessly to a laptop computer running proprietary software. The CurlSmart broom is highly accurate, as it is calibrated with the help of a force plate in the Kinesiology laboratory at Dalhousie’s main campus, under the direction of Professor Michel Ladouceur of Dalhousie’s School of Health and Human Performance. The accompanying photograph shows the author as the test subject in a calibration test of a PT-2 broom in the laboratory at Dalhousie, using the lab’s AMTI BP400600‐100 force plate, which has a 450kg capacity. Vertical force data, along with other readings from the force plate, were collected at 200 Hz (readings per second).
An example analysis from the CurlSmart broom is shown above, with the subject being a junior-aged male athlete from a competitive team. The mean stroke rate is 4.6 strokes per second over the entire shot, with a mean maximum force of 47.2 kilograms and a stroke average force of 27.5 kilograms. The oscillation of the force curve is a direct result of the sweeping motion, with the greatest force coming at the beginning of the push stroke, and the least at the beginning of the pull stroke when the brush is furthest from the player’s feet. Hence the athlete’s “force envelope”, or the difference between the peak and base (minimal) force through the trial, is 55.7 kg. Of particular interest in the graph is the periodic decrease in force every five or six strokes, which at times drops the sustained force (the “valleys” of the force curve) to near zero. This is entirely due to the player using a slider on his lead foot while brushing, rather than the recommended method using grippers on both feet. Any body weight applied to the slide leg to slide down the sheet of ice is, naturally, not able to be applied through the brush head into the ice. It is precisely this detailed, accurate feedback that makes an instrumented curling broom a “game changer” in the coaching of the sport.
This past November, I was fortunate to be asked by Helen Kontozopoulos to present a talk on instrumented curling brooms at the 2014 SmartWeek Toronto conference held at the University of Toronto. The slides from my talk can be viewed by clicking on the image below:
As stated earlier, one aspect of our research is to determine precisely what metrics from an instrumented curling broom should be captured, and what technology is required in order to make that possible (and with what error guarantees). A second aspect is to develop better coaching and instructional methods that can utilize the feedback from the broom in the most effective way for any athlete, regardless of where that athlete falls in the Canadian Curling Association’s Long-Term Athlete Development Model (LTAD). A third goal is to develop common nomenclature so that the outputs of various devices can be usefully compared. Finally, a fourth goal is to develop a set of normative data for a wide variety of athletes so that individual athletes can be compared against other athletes of similar characteristics.
Access to normative data is critically important in coaching, but until now the number of published results has been exceeding small. As one example, consider the results from Bradley  for five male and twelve female “elite” curlers:
|M/F||Avg Sweep Length||Avg Stroke Rate||Avg Total Work||Avg Heart Rate||Vertical Force|
|M||10.71 cm||4.32||1538 J||169||146.3 N|
|F||10.71 cm||3.81||663 J||164||81.7 N|
Bradley presents the Vertical Force results in Newtons (N), rather than in kilograms. Converting to kilograms (dividing by 9.8) yields a stroke average force of 14.92 kilograms for men, and a stroke average force of 8.33 kilograms for women. These figures pale in comparison to the junior-aged male athlete above, whose stroke average force was 27.5 kilograms. Similarly, Buckingham and Marmo [2,9] state that Olympic-calibre male athletes can achieve a peak force of approximately 450N, or 45.91kg. Again, this peak force is easily eclipsed by the junior athlete shown above, whose peak force is 56.0 kg. Our preliminary test results, using both the PT-2 commercial broom and the CurlSmart prototype, indicate that elite Canadian junior and University-aged athletes can routinely exceed these published results. Publishing a more wide-ranging study of athlete brushing performance, then, is a prerequisite to improving brushing technique through coaching.
We would like to thank Balance Plus for their generous sourcing of equipment for the Conestoga broom prototype; Conestoga students Brandon Davies and Aaron Schryver for their work on developing the Conestoga prototype; Curling Geek for their financial support of the project; John Newhook and his team at Dalhousie for numerous discussions and generous Maritime hospitality; TrainSmart for Curling of Halifax; Andrew Flemming of Canada Curling Tools; the coaches and athletes at Wilfrid Laurier University; and Gary Crossley and the Ontario Curling Council for sponsoring this work.
 Bradley, J. L. 2009. The sports science of curling: a practical review. Journal of Sports Science and Medicine 8:495-500.
 Buckingham, M.P., B. Marmo, and J. Blackford. 2006. Design and use of an instrumented curling brush. Proceedings for the Institution of Mechanical Engineers, Part L. Journal of Materials: Design and Application 220(4):199-205.
 Denny, M. 2002. Curling rock dynamics: towards a realistic model. Canadian Journal of Physics 80:1005-1014.
 Denny, M. 2003. Comment on “The Motion of a Curling Rock”. Canadian Journal of Physics 81(6): 877-881.
 Jensen, E. T., and M. R. A. Shegelski. 2004. The motion of curling rocks: experimental investigation and semi-phenomenological description. Canadian Journal of Physics 82:791-809.
 N. Maeno. 2010. Curl Mechanism of a Curling Stone on Ice Pebbles. Bulletin of Glaciological Research 28:1-6.
 N. Maeno. July 2013. Dynamics and curl ratio of a curling stone. Sports Engineering 17:33-41.
 Marmo, A. A., I. S. Farrow, M-P Buckingham, and J. R. Blackford. 2006. Frictional heat generated by sweeping in curling and its effects on ice friction. Proceedings of the Institution of Mechanical Engineers, Part L.: Journal of materials: Design and Applications 220:189-197.
 Marmo, B. A., M-P Buckingham, and J. R. Blackford. 2006. Optimising sweeping techniques for Olympic curlers. The Engineering of Sport 6(3):249-254.
 Nyberg, H., S. Hogmark, and S. Jacobson. 2012. Calculated trajectories of curling stones sliding under asymmetrical friction. Conference Paper from the 16th Nordic Symposium on Tribology pp. 12-15.
 Nyberg, H., S. Alfredson, S. Hogmark, and S. Jacobson. 2013. The asymmetrical friction mechanism that puts the curl in the curling stone. Wear 301:583-589.
 Penner, A. R. March 2001. The physics of sliding cylinders and curling rocks. American Journal of Physics 69(3):332-339.
 Shegelski, M. R. A. 2001. Maximizing the lateral motion of a curling rock. Canadian Journal of Physics 79:1117-1120.
 Shegelski, M. R. A. and R. Niebergall. 1999. The Motion of Rapidly Rotating Curling Rocks. Australian Journal of Physics 52:1025-1038.
 Shegelski, M. R. A. and M. Reid. 1999. Comment on: Curling rock dynamics – The motion of a curling rock: inertial vs. noninertial reference frames. Canadian Journal of Physics 77:903-922.
 Shegelski, M. R. A., R. Niebergall, and M. A. Walton. 1996. The motion of a curling rock. Canadian Journal of Physics 74:663-670.
 Shegelski, M. R. A., M. Reid, and R. Niebergall. 1999. The motion of rotating cylinders sliding on pebbled ice. Canadian Journal of Physics 77:847-862.
 Yanagi, H., K. Miyakoshi, M. Fukuoka, and N. Yamamoto. 2012. Development of Curling Brush for Measuring Force Exerted During Sweeping. Proceedings, 30th Annual Conference of Biomechanics in Sports, Melbourne, Australia, pp. 354-356.