NASA Cost Estimating Handbook Version 4.

Appendix K: Schedule Estimating Relationship (SER)

Development and Model Inputs
There is a fundamental relationship between project schedule behavior and project cost behavior.
Understanding this relationship is essential to the cost estimator. The following topics are described in
this appendix:

K.1. Background
K.2. NASA Experience and Usage
K.3. Project Schedule Analysis Methods
K.3.1. Parametric Schedule Estimating
K.3.2. Analogy Schedule Estimating
K.3.3. Engineering Build-Up/Grassroots Schedule Estimating
K.4. SER Data, Databases, and Models
K.4.1. NAFCOM Model
K.4.2. NICM Model
K.4.3. QuickCost Model
K.4.4. Schedule Estimating Relationship Risk Assessment (SERRA)
K.4.5. Schedule Management and Relationship Tool (SMART)
K.5. Schedule Data
K.6. SER Research
K.7. SER Example Using SERRA

K.1. Background
There is a fundamental relationship between project schedule behavior and project cost behavior.
Understanding this relationship is essential to the cost estimator. While this cost and schedule
relationship seems intuitive to those in project management, it is often difficult for an estimator to quantify
or model for the purposes of analysis. To complicate matters further, there is a distinction between the
relationship of cost and schedule and the correlation between cost growth and schedule growth. The
former (relationship of cost and schedule) is not always obvious, whereas the latter (correlation of cost
growth and schedule growth) is mostly self-evident. For example, it is widely observed in a multitude of
programs and projects that schedule growth usually leads to cost growth. In addition, the integrated
master schedule needs to correspond to cost estimates to ensure that enough resources can be applied
to activities to complete them within the expected duration. This should be done before the project
schedule is finalized so that the relation between accurate cost and schedule estimates can be verified.
Ascertaining specific values of correlations for a particular project is a challenge. As with Cost Estimating
Relationships (CERs), there are quantifiable relationships between the schedule and other project factors

Appendix K
SER Development and Model Inputs K-1 February 2015
 NASA Cost Estimating Handbook Version 4.0

and influences. Schedule Estimating Relationships (SERs) can likewise be used to estimate durations of
schedule events much like CERs are used to estimate a particular price or cost.
SER is a technique used to estimate schedule duration by connecting an established relationship with
one or more independent variables to the duration time of an event. If an independent variable (driver)
demonstrates a measurable relationship with schedule duration, an SER can be developed. While
relatively simple in concept, the ability to get accurate and meaningful data that can be used to quantify a
relationship between an independent variable and schedule duration can be difficult.
This appendix provides a basic overview of NASA schedule estimating. Additional information on the
subject is available in the NASA Schedule Management Handbook.1

K.2. NASA Experience and Usage

NASA’s SER experience has been confined to the project level (spacecraft, vehicle, and major element),
where analysis of aerospace projects moving from major milestones can be accomplished without the
influence of lower-level distortions. This high-level analysis has yielded statistical data and some
observable trends. A transition from study to implementation has begun recently with analysts using
SERs, schedule data, and models to complete schedule risk analyses on programs and projects.
While CERs can be directly used in project cost estimation, SERs do not have the same direct
applicability to schedule estimation as their cost counterparts. For example, while requirements affect
both cost and schedule, changes in requirements do not affect both cost and schedule estimates equally.
In many cases, this is primarily because of the immaturity of SER usage within NASA and the estimating
community. SERs, for the most part, are not as well publicized or used for schedule estimation. This
means that many analysts may not know that SERs exist or know whether they are applicable to their
particular program or project. Schedule estimation itself is also relatively new at NASA, and conducting a
risk analysis on schedule, where an SER would be used, is still not a common type of analysis performed
by programs or projects. Nevertheless, the new policies regarding high and low estimates of cost and
schedule, as well as a greater push to understand schedule, have led to an increase in the amount of
schedule analyses completed at the Agency.
CERs and SERs share similarities. Although CERs do not typically have schedule as an input, the same
factors usually drive project CERs and SERs. These factors could be technical, such as mass and power,
or they could be programmatic, such as funding stability and international partnerships. This speaks to
the fundamental and intuitive notion of the common relationship between cost and schedule within an
effort, since the same factors are often used to estimate both cost and duration.

K.3. Project Schedule Analysis Methods

Parametric, analogy, and bottom-up estimating methodologies used in cost estimating are also applicable
to schedule estimating. Estimating methods for schedule can be built using historical datasets,
comparison to analogy, or detailed grassroots analysis. Estimating schedule durations can be extremely
useful at both a high summary level or at the low activity level. The development of SERs, models, and
logically linked schedules is a key component of schedule estimating. The level of detail selected will
influence all of these key components. The selection of the schedule estimation method is tied to the
relationship between cost and data availability, estimate purpose, data maturity, and program maturity
levels. See Appendix C for more details.

K.3.1. Parametric Schedule Estimating

1 [Link]

Appendix K
SER Development and Model Inputs K-2 February 2015
 NASA Cost Estimating Handbook Version 4.0

Estimating schedule using a parametric approach involves the same fundamentals as estimating cost.
Schedule duration data and independent variables are collected to conduct data analysis and determine if
there are statistically significant relationships present to produce a SER. SERs can contain many of the
same independent variables as CERs but could also be based on different datasets, normalization
techniques, or analysis methods.

K.3.2. Analogy Schedule Estimating

The analogy method for schedule estimating focuses on comparing the estimated duration with actual
schedule data of similar missions. As with the analogy method for cost, careful consideration should be
given to selecting the analogies. Any adjustments, deviations, or differences should be clearly identified
and documented.

K.3.3. Engineering Build-Up/Grassroots Schedule Estimating

Using a detailed engineering build-up estimate to develop a schedule estimate is a common technique
employed by program managers all over the world. A highly detailed and logically linked schedule is the
standard product generated by this schedule estimation method. As with cost estimating, grassroots
estimating for schedule requires the same strong attention to detail to be successful. Analysts should
continue to be careful when differentiating between a build-up schedule estimate and a given detailed
schedule plan. Both may employ the engineering build-up/grassroots approach; however, there are
significant differences. The former reflects an attempt to capture the entire work effort to analyze
durations and the program plan. A build-up schedule estimate, similar to cost, is an attempt to predict the
actual (i.e., actual duration/actual finish date.) The latter reflects the result of a detailed project plan and
may contain significant constraints, optimism, or undocumented assumptions. The plan duration and plan
finish date from a given detailed schedule plan are attempts to organize future work with the goal of
delivering on time.
Collected data need to be normalized. Normalization involves analyzing the raw data and making
adjustments for consistency. The inconsistencies that may be found in a dataset include changes in dollar
values over time (inflation); learning or cost improvements for organizational efficiency; and, if more than
one unit is being produced, the effects of production rates on the dataset being analyzed.
When analyzing a dataset, normalization considerations should include adjustments for the cost
(currency, Base Year), size and weight, complexity or mission, frequency, and mission platform (crewed,
robotic).

K.4. SER Data, Databases, and Models

The underlying data that support the development and use of SERs are just as important as the data that
support the development and use of CERs. The challenge at NASA is that these data are not as well
recognized or available as similar cost data. Fortunately, recent efforts have begun collecting, organizing,
and using schedule data to develop SERs for NASA analysts. This will result in the creation of new
schedule estimating tools, as well as the expansion of existing cost models into the schedule estimating
arena. Existing cost estimating models and tools such as the NASA Air Force Cost Model (NAFCOM2)
also contain schedule data and SERs that can be used effectively, if the purpose and utility of the results
are understood by the analyst. Refer to Appendix E for a summary of Agency-provided, as well as
NASA-licensed models and tools. The following sections describe several of the SER-related tools used
by the NASA community.

2
See the NAFCOM version 2011 Help file, located under “Process-Based Schedule.”

Appendix K
SER Development and Model Inputs K-3 February 2015
 NASA Cost Estimating Handbook Version 4.0

K.4.1. NAFCOM Model

NAFCOM is perhaps the most widely used cost
modeling tool that can provide schedule output to the PCEC replacing NAFCOM
user. Three levels of schedules are generated by At the time of publication, NAFCOM is still
NAFCOM: system level by stage, stage level by heavily used by the NASA cost estimating
subsystem, and subsystem level by process. The term community. However, NAFCOM users are in
“stage” refers to a major element of a launch vehicle the process of transitioning to the Project Cost
estimate, as in a two-stage or three-stage launch Estimating Capability (PCEC), which contains
vehicle. Earth-orbiting and planetary missions typically additional information and tools. PCEC
do not have stages, so this level of the Work incorporates NAFCOM models, as well as
Breakdown Structure is not used. The process-based models developed by various NASA Centers
view inside of NAFCOM provides estimated average and directorates. Since most users at this time
schedules at the process level for each subsystem and are more familiar with NAFCOM, this
system integration element. The user can click on any handbook will continue to reference it. The
of these process bars and view a further breakdown of expectation is that the functionality provided to
the processes required to develop and produce the users by NAFCOM and referred to here will
hardware element. Summary schedules are also continue with PCEC. See Appendix E for
provided in NAFCOM at the major system element additional information on PCEC.
level, broken out by subsystem and system integration
element, and at the complete system level, broken out
by major system element. The schedule is estimated using the launch year (provided by the analyst on
the Global input screen) as the end date. The model uses a combination of cost and technical parameters
to estimate the schedule durations for each of the subsystems and system integration elements. In the
current model version, users cannot alter schedule durations or determine schedule penalties or
benefits—only average schedules are provided.
The NAFCOM-process-based schedule results are primarily useful as a crosscheck of the project
schedule assumptions across the major milestone events. The utility of these NAFCOM schedule
estimates and their results can be seen as a part of the wider context of schedule metrics that provide
comparison to project planning and assumptions. Since the results provided by NAFCOM are average
schedules, they are best used as a crosscheck at a high level against the program plan. This is much like
other professional scheduling rules of thumb anchored by technical experience. The schedule metric
informs project managers about how their baseline plan compares to the historical data at a high level.
Schedule metrics, such as those from NAFCOM, can be used to compare against the results of project
planning efforts (outputs), as well as assist technical experts during the buildup of an initial estimate.
NAFCOM also provides a schedule database that can be used to better understand the schedule data for
particular programs and projects. Figure K-1 provides an example schedule output from the NAFCOM
model at the subsystem level.

Appendix K
SER Development and Model Inputs K-4 February 2015
 NASA Cost Estimating Handbook Version 4.0

ATP PDR CDR Delivery

Preliminary Design
Detailed Design
Fab & Assembly
Integration & Test

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110

Figure K-1. An Example Schedule Output from NAFCOM at the Subsystem Level
Questions about NAFCOM should be directed to Andy Prince at [Link]@[Link].

K.4.2. NICM Model

Another NASA modeling tool that contains schedule data and SERs is the NASA Instrument Cost Model
(NICM).3 The NICM, which is available via the ONCE Model Portal at [Link], focuses
specifically on instrument estimation and contains a large database of many different types of
instrumentation. This database includes schedule data, and there is a component within NICM for
estimating schedule duration using SERs. The NICM approach to calculating duration from SERs is
unique in that cost is an input to the SER equation. In this way, NICM SERs establish a functional link
between the calculated cost of an instrument and its schedule duration. In addition to utilizing Cost As an
Independent Variable (CAIV), NICM relies on the mission type and instrument subtype in the SER
equation. Figure K-2 shows the NICM model schedule equation.

Figure K-2. NICM SER Information

Analysts with questions about using the NICM for SER capability should contact Joe Mrozinski of the
NICM development team at the Jet Propulsion Laboratory (JPL), at jmrozins@[Link].

3 NICM version VI, Jet Propulsion Laboratory, May, 2014. For more information, see Appendix E.

Appendix K
SER Development and Model Inputs K-5 February 2015
 NASA Cost Estimating Handbook Version 4.0

K.4.3. QuickCost Model

The QuickCost4 model developed by Joe Hamaker also contains a schedule database and SERs. The
QuickCost schedule database contains duration data on satellites, engines, module and transfer vehicles,
and experimental vehicles (X-vehicles). SERs are available primarily for satellites. QuickCost will
calculate duration for X-vehicles and module and transfer vehicles, but this relies upon the SER
developed for satellites with additional calibration. QuickCost SERs focus on providing duration results for
Phases C and D, omitting duration for Phases A and B. Analysts seeking to use QuickCost for its SER
capability should contact Hamaker for more information.5

K.4.4. Schedule Estimating Relationship Risk Assessment (SERRA)

The SERRA modeling tool also contains a database and SERs that can be used to conduct schedule risk
analysis.6 The SERRA SERs leverage the NAFCOM database of technical dependent factors with the
explicit intent of generating NAFCOM-like SERs. SERRA can provide SERs and data on planetary, Earth-
orbiting, and launch vehicle/human-exploration missions. SERRA was developed for NASA by SAIC and
contains data on approximately 80 past projects. The available SERs can provide results for both Phase
B/C and D and can provide a good crosscheck at a system level.
SERRA also provides schedule risk analysis through two components: technical uncertainty of inputs to
the SER, and the error inherent in the SER regression. The underlying statistics of the SERRA SERs and
additional information can be provided by CAD. The SERRA model is available at no cost for NASA
Government distribution through the NASA Cost Analysis Division (CAD). Questions about SERRA
should be directed to CAD at hq-cad@[Link]. Sample schedule inputs and outputs for SERRA
are provided in Figure K-3 along with a tabular output shown in Figure K-4.

Figure K-3. SERRA Schedule Model Inputs

4 QuickCost 5.0, Joseph Hamaker, Ph.D., 2011.

5 jhamaker@[Link]. For more information on QuickCost, see Appendix E.
6 SERRA User Guide, SAIC, August 2010. For more information about SERRA, see Appendix E.

Appendix K
SER Development and Model Inputs K-6 February 2015
 NASA Cost Estimating Handbook Version 4.0

Figure K-4. SERRA Schedule Model Tabular Outputs

K.4.5. Schedule Management and Relationship Tool (SMART)

The Schedule Management and Relationship Tool (SMART) combines analogy-based and parametric
methods in a schedule estimating tool for unmanned spacecraft projects. The tool utilizes high-level
technical and programmatic characteristics to determine a spacecraft’s likely development schedule
duration. It incorporates several previously developed third-party schedule estimating relationships
(SERs) for comparison. SMART was developed by SpaceWorks Enterprises, Inc. (SEI) and is available
via the NASA ONCE Model Portal at [Link].

K.5. Schedule Data

Raw schedule data used to develop SERs can be found in various locations. The schedule data are
spread across the Agency, and analysts can expect to expend some effort during data collection.
Fortunately, recent efforts to consolidate and organize the data have made this job somewhat easier. The
Cost Analysis Data Requirement (CADRe) data and the One NASA Cost Engineering (ONCE) database
are excellent data sources for collecting duration information, at [Link]. Examining prior
research from the Aerospace Corporation as well as the Air Force Cost Analysis Agency (AFCAA) and
the National Reconnaissance Office (NRO) can also provide the analyst with previously created schedule
databases for SER development.

K.6. SER Research

In 2009 and 2010, NASA CAD commissioned MCR, LLC, to enhance and upgrade the database of
satellite and payload schedules, focusing on the time required to move from one development milestone
to the next. This work builds upon a previous database developed under a joint AFCAA/NRO task
organized by the NRO Cost Analysis Improvement Group (CAIG). This research resulted in the creation
of many individual SERs and a large database of schedule information for the Department of Defense
(DOD), NASA, and NRO missions.
The resulting unclassified spacecraft database for satellite SER development contains 247 data records
and is segregated by mission type. The resulting unclassified payload database contains 519 data
records and is also segregated by mission type.7 The data are further segregated by the various

7 NRO Cost Group (NCG) Data Books—an ongoing collection from Government and industry, 1995 to present.

Appendix K
SER Development and Model Inputs K-7 February 2015
 NASA Cost Estimating Handbook Version 4.0

hardware vendors and identified by agency sponsor (e.g., DOD, NRO, or NASA). Parameters for mission
design life (DL), end of life (EOL) date, and failure cause are included in the database to allow additional
analysis of actual versus estimated plots that may suggest a relationship between the probability of
premature failure and schedules that are less than the estimated schedules predicted by the SERs.
The research itself developed SERs for the time between spacecraft and instrument development
milestones, including Authorization to Proceed (ATP), Preliminary Design Review (PDR), Critical Design
Review (CDR), Integration and Test (I&T), Payload Delivery (DEL), First Launch Availability (FLA), and
others as supported by the data. Figures K-5 and K-6 summarize the spacecraft SER and instrument
SER types, respectively.

Figure K-5. NASA Spacecraft SER Types Developed by MCR

Figure K-6. Instrument SER Types Developed by MCR

The SERs were developed using three statistical methods: (1) Ordinary Least Squares (OLS), (2)
Minimum Percentage Error–Zero Percentage Bias (MPE-ZPB or ZMPE), and (3) Errors in Variables (EIV)
regression.
The resulting SERs’ goodness-of-fit statistics that were developed under this task, in general, are poorer
fit statistics (as expressed by error residuals, degrees of freedom, and R 2) than would be found in CERs.
The criteria for selecting recommended SERs were thus relaxed for this research to include SERs with a
standard error of 40 percent and R2 at or above 50 percent.
The two primary SER drivers that were collected and used in SER development were dry mass and
beginning of life (BOL) power. The primary driver of schedule is the size of the program as measured by
complexity. Both dry mass and BOL power are often used as proxies for measuring this complexity. In
general, the higher a spacecraft mass and the more power it requires, the more complex it is. Another
driver that is often seen in SERs is DL. Longer DL often necessitates more testing to ensure that the
spacecraft lasts as long as advertised. In order to consider the “newness” of the spacecraft, an indicator
variable for block change/new design was also collected in the research. This allows the analyst to
distinguish from follow-on spacecraft and spacecraft undergoing major changes that could result in

Appendix K
SER Development and Model Inputs K-8 February 2015
 NASA Cost Estimating Handbook Version 4.0

significant design challenges. Two additional metrics were collected to capture the amount of diversity
present in the spacecraft development process: the number of mission types and the number of distinct
payloads.
Quality metrics are calculated for each SER to provide the user with an understanding of the likely
accuracy of estimates based on that SER. These quality metrics depend on the values of the drivers in
the database alone and not on any probabilistic assessment of the expected growth of the drivers of the
particular SER involved. Therefore, during a typical risk analysis, the user should consider probable
growth in the drivers, which can help account for estimating the a posteriori value of schedule duration
using a priori values for schedule drivers.
Figure K-7 shows an example SER for ATP to FLA for a planetary spacecraft using ZMPE.
The complete research report, results, and Excel database are available from CAD upon request.

Figure K-7. SER for ATP to FLA of Planetary Spacecraft Using ZMPE

K.7. SER Example Using SERRA

The following example shows how an analyst can use programmatic data and a SER to estimate
schedule duration. In this example, the analyst is tasked to perform a parametric schedule estimate. The
analyst has reviewed preliminary program summaries and selected the planetary spacecraft mission type
from SERRA since these most closely match the current program.
The analyst collects the following technical and programmatic data from the Project Office8,9:
• Mission Type = planetary spacecraft
• Schedule Start (StartYr) = 2011 (schedule start Base Year is 1960)
• Engineering Management (StreamEM) = 25 (few design changes, Skunk-Works approach)
• Satellite Dry Weight (DryWt) = 1,800 pounds
• Maximum Data Rate (MaxData) = 256 kilobytes per second

8
The model variable names are provided in parentheses next to the input parameter name.
9
Both engineering management and funding availability are based on scales of 0 to 100.

Appendix K
SER Development and Model Inputs K-9 February 2015
 NASA Cost Estimating Handbook Version 4.0

• Number of Instruments (NumInst) = 8

• Funding Availability (FundAvail) = 50 (some infrequent delays possible)
• Mission Design Life (DesignLife) = 24 months from launch to end of mission (includes transit
time)
• Power Source: Radioisotope Thermoelectric Generator (RTG); (1 = no RTG, 0 = RTG)
The Engineering Management input describes the level of design changes, experience of the design
team, and environment of the design effort, as follows:
• 0—Minimum design changes with design team, making maximum use of highly efficient Skunk
Works approach with Integrated Product Teams (IPTs), rapid prototyping, design to cost, etc.
• 25—Few design changes using a highly efficient Skunk Works approach with IPTs, rapid
prototyping, design to cost, etc.
• 50—Moderate design changes with an application of advanced design methods, including
concurrent engineering, tailored specifications, minimum reporting, etc.
• 75—Dedicated design team dependent on some technology advances that encounter significant
requirements changes
• 100—Distributed design team dependent upon major technology advances and experiencing
frequent major requirements changes
The Funding Availability input reflects the appropriate anticipated funding availability. Choices for Funding
Availability include the following:
• 25—Funding is assured; no delays
• 50—Some infrequent delays possible
• 75—Funding is constrained; delays likely
The appropriate equations in SERRA for estimating planetary mission schedule phases are as follows:
Schedule Duration (ATP to CDR)

Schedule Duration (CDR to Delivery)

It is important to realize for log-transformed regression that a “no” input translates to a value of 1.0 and a
“yes” input translates to a value of 2.71828. StreamEM is a yes/no input based on a streamlined engineering
management approach (yes, in this case). Also for these SERs, the input for the year of schedule start is
normalized by subtracting 1960 from the year of start. The RTG input is yes/no as well, and since this
spacecraft is not powered by an RTG, the value input in the SER equation is 1, regardless of what other
power source might be used.
The SER equations with the appropriate programmatic inputs collected above are:
Schedule Duration (ATP to CDR)

Appendix K
SER Development and Model Inputs K-10 February 2015
 NASA Cost Estimating Handbook Version 4.0

Schedule Duration (CDR to Delivery)

Thus, the estimate development phase (ATP to CDR) duration is 21 months, and the estimated
manufacturing phase (CDR to delivery) duration is 27 months.

Appendix K
SER Development and Model Inputs K-11 February 2015