Everything you need to know about Zwift Trainer Difficulty
Updated: Feb 28
Zwift Trainer Difficulty is a misunderstood concept. This article provides a technical explanation of how adjusting Trainer Difficulty impacts rider effort level and experience and answers fundamental questions:
Is it cheating?
Does it make a rider go faster?
What effect does Trainer Difficulty have on the avatar “performance”?
What are the benefits of the Trainer Difficulty setting?
Zwift provides the ability to adjust what is termed Trainer Difficulty. This is done in the settings accessed from the Menu button. The default setting is 50% and it can be adjusted from 0% to 100%.
Several theories are circulating that are supposedly verified by Zwift, but the algorithm is the applications developer’s intellectual property, and they are not going to reveal their secrets. Furthermore, Trainer Difficulty is a misleading term and may better be expressed as Trainer Experience.
The cyclist must generate power to overcome gravity resistance, rolling resistance, and wind resistance as shown in Figure 1.
Two key concepts need to be stated:
Smart trainers can only control resistance.
The Zwift app simulates rolling resistance, wind resistance (drag), and gradient by varying the resistance applied by the smart trainer.
Figure 1: Image courtesy of Bergfreunde
The change in resistance is achieved by applying a variable voltage to an electromagnet. As the power applied to the electromagnet is increased the more power must be applied by the rider to maintain speed.
Figure 2. illustrates the forces the cyclist encounters on an incline (gradient).
Figure 2: Fᵣ = mgsinθ + μmgcosθ
The standard mechanical Resistive Force formula includes gradient, frictional (rolling), and air resistance(drag). Assuming a steady-state speed the equation appears as
Fᵣ = mgsinθ + μmgcosθ + ½CρAV²
m = mass of cyclist and bicycle (kg)
g = gravity (9.81 m/s2)
θ = gradient (in degrees)
μ = coefficient of rolling friction (usually 0.0022 to 0.0005 for rubber bicycle tires)
C = coefficient of drag
A = area of the object (m2)
P = density of air (kg/m3)
The major component of resistive force Fr is attributable to the gradient component of the equation (mgsinθ). On flat terrain at high speed, the drag force becomes the major component of the resistance.
Power to Speed relationship
Zwift users generally understand that there is a relationship between velocity (or speed - as the moving cyclist is always going forward) and power and mass:
V α P/M
(α is proportional to)
V is velocity (m/s)
W is power in Watts
M is mass in Kg
One normally writes this as
Speed α W/Kg
The Zwift user must apply sufficient power to the pedals to overcome the Resistive Force components. The effective Power, Pe is therefore
Pₑ = Pₐ - PFr
Where Pₑ is Effective Power, Pₐ is Power Applied, and PFr is the power required to overcome the Resistive Force, Fr.
The equation for speed is more correctly indicated as
V α Pₑ/M (Effective Power/Rider Mass).
A 65 kg rider with a Functional Threshold Power (FTP) of 145W, traveling at 30km/hr and producing 130 W or 2W/kg on a flat terrain will experience a rolling resistance of -0.3W/kg. Ignoring drag, the Effective Power will be 1.7W/kg. This is easily manageable for the rider’s ability and sustainable for a long period.
The relationship between Velocity and Effective Power (Pₑ) is not changed by the gradient. This can be demonstrated by the following example:
The same rider as above, on a gradient of 15% (8.5o)will experience an additional -1.5W/kg resistive force due to gradient. The Effective Power at the same Power Applied will be reduced to 0.2 W/kg and the Velocity will be proportional to this amount i.e. almost no forward motion.
The rider should be able to sustain a power output of 110% of 145W (160W or 2.46W/kg) for 20 minutes so has the potential to apply only an additional 0.46W/kg resulting in an Effective Power of 0.66W/kg and applied proportionally to the velocity will achieve some forward speed.
Note that the values for rolling resistance and gradient resistance above and in the graph below are not meant to be accurate but only to be representative of the Speed/Force relationship.
Figure 3: Effective Power/Mass
The graph shows the initial rolling resistance component (μmgcosθ) as there is a minimum power requirement before movement is possible. Once the minimum power is applied thereafter there is a linear relationship between Effective power (not to be confused with Power Applied) and Velocity.
Impact of Drag
Drag is important both in the real world and in the Zwift world because it produces a resistive force that has a non-linear impact on speed.
As stated above the equation for Force attributable to drag is F = ½CρAV²
Zwift calculates A (area) from rider height.
C (drag coefficient) and ρ (air density) can be considered as constants.
The simulated non-linear resistive force is therefore a function of CV²/2 (due to the square of the Velocity). As virtual speed increases so does the resistive force increase. The Speed/Drag (force) graph below demonstrates the effect only and is not to be considered actual figures:
Figure 4: Drag/Velocity
At a speed of 30 km/h the power attributable to drag is 22.5W whereas at 60 km/k it is 90W.
Therefore, resistive force attributable to drag is relative to speed only and not the angle. The resistive force is attributable to drag decreases on a virtual climb because of the riders’ speed reduction and increases on a virtual descent because of speed increase. Drag is also the reason that a rider’s top-end speed is limited.
Impact of Inertia
Inertia is the resistance of any physical object to any change in its velocity. There is a mechanical inertia effect generated by the flywheel. Zwift measures this and compensates for this in its algorithm. It is Zwift that creates a simulation of the resistance required to accelerate the virtual rider (avatar) from an initial steady-state speed to a higher steady-state speed. It is not the smart trainer flywheel.
The unit for acceleration is meter per second squared = m/s².
Force (N) = mass (kg) × acceleration (m/s²).
Zwift simulates the mass (kg) incorporating data obtained from the smart trainer during the calibration spin down. The avatar acceleration is based on simulated inertia and not only the smart trainer flywheel inertia. For acceleration from 20km/hr to 30mkm/hr the rider must apply the same amount of power for the same amount of time irrespective of Trainer Difficulty.
Trainer Difficulty (or Trainer Experience)
At 15% gradient and at 50% trainer difficulty the total resistive force is substantial. Many Zwift riders may not be capable of sustaining the required minimal power in the easiest gear (due to the bicycle gear ratio) to climb the grade. This virtual simulation corresponds to that of the real world, and the solution in the real world would be to change the gear ratio but this is not practical in a virtual training environment. The Zwift app provides a solution that makes 15% climbs accessible to all Zwift users.
Zwift states that a 50% Trainer Difficulty setting simulates 50% of the gradient. Converting a 15% gradient to degrees results in an θ of 8.5° and 50% of 8.5° results in a θ of 4.25°. However, if one considers the gradient formula (mgsinθ) the difference between sinθ (where θ = 4.25°) and 50% of sinθ (where θ = 8.5°) is insignificant (sin 4.25o = 0.074108 vs 1/2sin 8.5o = 0.073905). A 50% Trainer Difficulty setting can therefore be derived by simply applying 50% to the gradient force calculation. Even though Zwift explains the principle of Trainer Difficulty as a reduction in gradient this is probably to avoid complexity.
The result of adjusting Trainer Difficulty is that Zwift provides the user a selectable percentage of Gradient Resistive Force. Depending on their ability the user can select a gear and maintain a cadence that overcomes the reduced resistance. However, this is not “virtual gearing” as the effect on the riders’ experience is because of a change in resistance (0 to 100%). Users that can manage a 100% Trainer Difficulty will be provided the full experience while less powerful riders can select a lower Trainer Difficulty.
It should be noted that the rider's power output ability is unchanged as well as the relationship to speed in the virtual world. If the Pₐ - PFr produces a very small Pe the virtual speed will be correspondingly small.
The effect of Trainer Difficulty on descending needs to be considered. The mgsinθ component of the gradient force calculation becomes negative on a descent.
· Pₑ = Pₐ - (-PFr) resulting in Pₑ = Pₐ + PFr
On a descent, with 50% Trainer Difficulty, the user experiences 50% of the benefit of the gradient i.e. 50% of 15% = 7.5%. The Effective Power (Pₑ) is reduced, and the rider may need to apply power to maintain speed. There is no speed advantage over riders with a higher Trainer Difficulty.
At 0% Trainer Difficulty the gradient force (Fᵣ) is completely removed, and the rider experience will be that of a flat road, and only the rolling and drag resistive force components applied by Zwift.
“Normalized Power® (NP) is a power averaging method, measured in watts, used to compensate for changes in ride conditions for a more accurate depiction of power expenditure.” (Garmin Support, 2021).
Adjusting Trainer Difficulty will have a significant effect on NP since the variation due to virtual terrain is reduced. For non-ERG training rides, this needs to be considered by the athlete or athletes’ coach. Should an athlete be training for a flat time trial a lower training difficulty may be appropriate but for an athlete training for a hilly Gran Fondo a higher tolerable Trainer Difficulty may be advantageous.
Training Stress Score
“Training Stress Score (TSS) is a composite number that takes into account the duration and intensity of a workout to arrive at a single estimate of the overall training load and physiological stress created by that training session.” (TrainingPeaks Help Center, 2021)
In the same way as adjusting Trainer Difficulty influences NP, it will also impact TSS. There is a natural tendency to apply more power when gradient increases and by reducing Trainer Difficulty these natural variations are removed. Riding a flat route in Zwift with a low Trainer Difficulty will produce a lower TSS score (but significantly higher virtual mileage) than the same route with a higher Trainer Difficulty.
Cheating in Zwift
It is important to enter personal data accurately to have a realistic ride experience in Zwift. Zwift uses a) rider height to calculate drag, and b) rider weight, for the gradient resistance calculation. If one does not use Zwift for racing or Strava KOM’s, then natural plus/minus variances should not be too much of a concern. Fitbit or Withings weight data can also be automatically uploaded into Zwift.
Cheaters can be easily identified. Rider weight can be calculated from the Zwift pictures on Strava by dividing the instantaneous power by the instantaneous W/kg. Significant changes, such as from 56kg to 50kg, or an improbable low weight is probably an indication of manipulation to gain a racing advantage.
Adjusting Trainer Difficulty cannot be used for cheating in a race. The avatar speed and performance are a function of Effective Power. The same Average Power over distance is required for any Trainer Difficulty, and the same applied W/kg is required for the same virtual speed. Therefore, the time achieved on a climb (e.g. Alpe du Zwift) or flat route (e.g. Tempus Fugit), is the same irrespective of the Trainer Difficulty. However, although it is not virtual gearing, the effect of reducing Trainer Difficulty is similar to having more suitable (lower) gearing.
Racing and Training in Zwift
Although Trainer Difficulty cannot be used to cheat, an adjustment can provide a completely legal advantage in the same way as gear selection is in the real world. Depending on the bike gearing, rider strength, and course profile the optimal Trainer Difficulty can be selected. An example would be on a course that includes gradients not exceeding 10%. The rider can select a Trainer Difficulty that allows a cadence of (no less than) 80 rpm in the easiest gear.
For training purposes, it has already been mentioned that Trainer Difficulty can be changed according to the specific athlete focus. It can also be changed to maximize the experience and enjoyment. Two examples are using 100% Trainer Difficulty on a relatively flat route or using 40% Trainer Difficulty on a mountainous route. In the first instance, the rolling terrain can provide a pleasant variation in effort whereas in the second instance allows the user to experience the challenge of the climb. There should be no concerns that a lower Trainer Difficulty reduces the achievement.
The Zwift Trainer Difficulty (TD) setting:
1. Reduces the resistance applied by the smart trainer.
2. Allows the rider to maintain a sustainable cadence.
3. Uses the same power to speed relationship as in the virtual world.
4. Provides no virtual world speed advantage.
5. Is not virtual gearing but can be perceived as such.
6. Enables riders with lower ability to experience virtual climbs that would be beyond their capability with their real-world equipment and gear ratios.
Richard John Cound 25th January 2021
For their review and contributions:
· Theo Grier
· Jim Landon
· Pauline Cound